MICRO-FABRICATED IMPLANTABLE HYBRID
NEURAL INTERFACES FOR OPTOGENETICS
By
KI YONG KWON

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Electrical and Computer Engineering - Doctor of Philosophy
2014

ABSTRACT
MICRO-FABRICATED IMPLANTABLE HYBRID NEURAL
INTERFACES FOR OPTOGENETICS
By
KI YONG KWON
Neural interfaces are a direct communication pathway between nervous systems and
external environment enabling promising clinical treatments for neurological disease and events
such as spinal cord injury, stroke, and traumatic major amputations. Existing artificial neural
interfaces mainly use electrical signals to evoke sensation in the central- and peripheral- nervous
systems (CNS and PNS respectively), and these electrical systems have become a powerful tool
for decades. However, its limitations demand improved technology. Recent developments in
optogenetics have demonstrated the ability to target specific types of neurons with submillisecond temporal precision via direct optical stimulation of genetically modified neurons in
the brain.
Current optogenetics-based stimulation interfaces mainly use laser- or light emitting
diode (LED)-coupled optical fiber, micro-LEDs (μ-LEDs) array, and a laser beam focused
through a microscope as their light sources. For experiments with freely behaving subjects,
however, only limited light delivery methods are available. These systems’ poor spatial
resolution limits their functionality, and the tethered optical fiber greatly restricts subjects’
natural behaviors.
To address these limitations, a series of hybrid neural interfaces based on a polymerbased flexible μ-LEDs array has been developed, and these arrays are designed to provide a
unique solution to the current demand for multichannel, bi-directional neural interface devices.

Depending on interface location, there are three different neural interfaces: Opto-μECoG array
for epidural stimulation, three dimension (3-D) waveguide array for deep cortical stimulation,
and slanted 3-D waveguide array for multi cortical layers stimulation. The Opto-μECoG array
contains a transparent micro- electrocorticogram (μECoG) electrode array with integrated μ-LED
bare dies. To further improve the spatial and depth resolutions of the optical stimulation via the
μ-LED, two types of 3-D multi-LED arrays were developed by coupling micro-scale optical
waveguides with LED chips using a polymer-based microfabrication technology. Integration of
individually addressable μ-LED chips with varying-length microneedles enables precise light
delivery to target neurons in specific cortical layers.
To demonstrate the wireless capability of the arrays, this author and colleagues proposed
a wireless-powered, multichannel optrodes array that is capable of simultaneous light stimulation
and electrical neural recording.
Based on the development of hybrid neural interfaces, an optogenetics-based cortical
visual prosthesis has been designed for treating all forms of blindness. With the hybrid neural
interfaces, the core hypothesis that epidural optical stimulation of microbial opsin expressing
neurons in the primary visual cortex (V1) can induce phosphine, artificial visual sensation, and
perception, has been tested.

Copyright by
KI YONG KWON
2014

In dedication to my parents for making me be who I am, my wife for being the woman of my life,
and my son for being a greatest gift of my life

v

TABLE OF CONTENTS

LIST OF TABLES..................................................................................................................... ix
LIST OF FIGURES..................................................................................................................... x
CHAPTER 1
A hybrid neural interface with integrated LEDs for optogenetics.............................................. 1
CHAPTER 2
A review of optogenetics and light delivery methods................................................................. 9
2.1 Introduction............................................................................................................... 9
2.2 Light sensitive proteins............................................................................................. 9
2.2.1 Excitation................................................................................................. 10
2.2.2 Inhibition.................................................................................................. 12
2.2.3 Bi-stable................................................................................................... 14
2.2.4 Signaling.................................................................................................. 15
2.3 Light delivery methods........................................................................................... 16
2.3.1 Waveguide-based systems....................................................................... 16
2.3.2 Light distribution systems........................................................................ 19
2.3.3 Optrodes................................................................................................... 20
2.3.4 Polymer-based systems............................................................................ 22
2.4 Specific studies based on optogenetics and their methods..................................... 24
CHAPTER 3
Opto-μECoG array.................................................................................................................... 27
3.1 Motivation............................................................................................................... 27
3.2 Design..................................................................................................................... 29
3.2.1 A multichannel, bi-directional system..................................................... 29
3.2.2 Maximum target cortical area of the optical stimulation......................... 30
3.2.3 Maximum spatial resolution of optical stimulation and electrical
recording............................................................................................................ 30
3.2.4 Biocompatibility and minimum invasiveness.......................................... 30
3.3 Fabrication............................................................................................................... 31
3.3.1 Fabrication of the μECoG electrode array............................................... 32
3.3.2 Fabrication of a PDMS stamp.................................................................. 32
3.3.3 Substrate fabrication and assembly.......................................................... 33
3.4 Results and discussions........................................................................................... 33
3.4.1 Optical properties..................................................................................... 34
3.4.2 Electrical properties.................................................................................. 35
3.4.3 Thermal properties................................................................................... 36
3.4.4 Soak testing.............................................................................................. 38
vi

3.4.5 In vivo animal experiments....................................................................... 38
3.5 Conclusion............................................................................................................... 43
CHAPTER 4
Three-dimensional polymer waveguide array for optogenetics................................................ 45
4.1 Motivation............................................................................................................... 45
4.2 Design...................................................................................................................... 48
4.3 Fabrication............................................................................................................... 51
4.3.1 Multi-LEDs array fabrication................................................................... 51
4.3.2 A 3-D μ-waveguides array fabrication and assembly.............................. 52
4.3.3 A slanted μ-waveguides array fabrication and assembly......................... 52
4.4 Results..................................................................................................................... 53
4.4.1 Optical properties..................................................................................... 54
4.4.2 Electrical properties.................................................................................. 57
4.4.3 Mechanical properties.............................................................................. 57
4.4.4 Long-term biocompatibility..................................................................... 61
4.5 Conclusion............................................................................................................... 62
CHAPTER 5
Varying-length polymer microneedle arrays fabricated by droplet backside exposure............ 64
5.1 Motivation............................................................................................................... 64
5.2 Methods................................................................................................................... 69
5.2.1 Principle of the DBE................................................................................ 69
5.2.2 Volume shrinkage and gravity effect....................................................... 72
5.3 Fabrication methods................................................................................................ 73
5.3.1 Basic DBE fabrication process................................................................. 73
5.3.2 A slanted hollow microneedle array fabrication process......................... 75
5.4 Results..................................................................................................................... 77
5.5 Conclusion............................................................................................................... 78
CHAPTER 6
Development of a cortically based hybrid visual prosthesis..................................................... 81
6.1 Cortically based visual prosthesis........................................................................... 81
6.2 Three major tasks for development of the hybrid visual prosthesis........................ 83
6.2.1 Stimulus-driven ECoG data analysis........................................................ 83
6.2.2 Chronic versions of the hybrid neural interfaces..................................... 83
6.2.3 Design of a behavioral animal experiment............................................... 84
6.3 Investigation of phase-locked neuronal oscillation................................................. 86
6.4 Phase synchrony estimations................................................................................... 87
6.5 In vivo recordings.................................................................................................... 88
6.6 Results..................................................................................................................... 90
6.7 Conclusion............................................................................................................... 93
CHAPTER 7
Chronically implantable wireless hybrid neural interfaces....................................................... 95
7.1 A wireless slated optrode array............................................................................... 95

vii

7.2 Methods................................................................................................................... 98
7.2.1 Slanted waveguide array.......................................................................... 98
7.2.2 Switch-capacitor stimulation (SCS)......................................................... 99
7.3 Fabrication............................................................................................................. 102
7.4 Results................................................................................................................... 104
7.4.1 In vivo animal experiments.................................................................... 104
7.4.2 Hermetic sealing of chronic implants..................................................... 109
7.5 Conclusion............................................................................................................. 112
CHAPTER 8
Behavioral animal experiment................................................................................................ 113
8.1 Motivation............................................................................................................. 113
8.2 A simple battery-powered wireless optical neuromodulator for freely behaving
mice............................................................................................................................. 114
8.2.1 Magnetic-switch based controller.......................................................... 116
8.2.2 IR-switch based controller..................................................................... 117
8.2.3 IR-MCU based controller....................................................................... 118
8.3 Behavioral experiment for tinnitus detection in a mouse model........................... 119
8.3.1 GPIAS method....................................................................................... 120
8.3.2 The modified GPIAS method................................................................. 120
8.4 Results................................................................................................................... 121
8.4.1 Characterization of the battery-powered wireless neuromodulator....... 121
8.4.2 In vivo animal experiment ..................................................................... 123
8.4.3 Behavioral animal experiment............................................................... 124
8.5 Conclusion............................................................................................................ 125
CHAPTER 9
Conclusion............................................................................................................................... 126
BIBLIOGRAPHY................................................................................................................... 129

viii

LIST OF TABLES

Table 1.

Known single-component optogenetics tools for excitatory................................ 11

Table 2.

Known single-component optogenetics tools for inhibitory................................ 13

Table 3.

Known single-component optogenetics tools for bi-stable excitatory................. 14

Table 4.

Known single-component optogenetics tools for bio signaling........................... 15

ix

LIST OF FIGURES

Figure 1-1.

Types of neurophysiological signals and their temporal and spatial resolution
[14]...................................................................................................................... 2

Figure 1-2.

Different optogenetics-based hybrid neural interfaces for optical stimulation
and electrical recordings of cortical neurons: (a) a conceptual illustration of the
optical neuromodulator integrated with LEDs (b) Opto-μECoG array (c) 3-D
multi-LED arrays coupled with micro optical waveguides (d) and slanted
waveguides.......................................................................................................... 4

Figure 1-3.

The overall design of the wireless optrode neural interface................................ 6

Figure 2-1.

Basic properties of single-component optogenetics tool adapted from [39]..... 10

Figure 2-2.

Waveguide-based light delivery methods: (a) laser- or LED-coupled optical
fibers and the light delivery through an implanted cannula [22], (b) implantable
three-dimensional sets of silicon oxynitride waveguides [79], and (c)
multipoint-emitting optical fibers [81].............................................................. 17

Figure 2-3.

Optrode-based light delivery methods: (a) multi-array silicon probes with
integrated optical fiber [74], (b) a Utah-style multi-array with an optrode [86],
(c) a neural probe with monolithically integrated electrodes and waveguide
[88], and a tungsten microelectrode-based optrode that encloses optical fibers
within its insulation glass [90].......................................................................... 21

Figure 2-4.

Polymer-based light delivery methods: (a) the chip level integration technology
[96], (b) a polymer-based single optrode shaft integrated with microfluidic
channel [89], and (c) a flexible optrode with integrated micro-LEDs using
micro-contact printing process [99].................................................................. 22

Figure 3-1.

A concept diagram of BMI system with an Opto-μECoG array....................... 28

Figure 3-2.

Concept illustration of the proposed Opto-μECoG array.................................. 29

Figure 3-3.

Opto-μECoG array fabrication process: (a) processes of μECoG electrode array
fabrication; (b) PDMS stamp fabrication; and (c) LED array substrate
fabrication and integration with the μECoG electrodes.................................... 31

Figure 3-4.

A fabricated Opto-μECoG array and light scattering test: (a) ITO electrode
array, (b) the assembled Opto-μECoG array, and (c) zoomed-in image of
individual modules. (d) LED array before hot water reflow (left) and after the
x

reflow (right). (e) Light scattering property of the µ-LED chip was estimated in
brain tissue and (f) a normalized relative light intensity was estimated. Both
color and the height (z-axis) indicate blue light intensity................................. 34
Figure 3-5.

Optical and electrical properties: (a) visible transmission spectra of ITO film
and (b) I-V characteristic of a μ-LED chip....................................................... 35

Figure 3-6.

Thermal properties: Heat dynamics of the Opto-μECoG array at various input
voltages (2.7 V-3.2 V) for 10 ms (left) and 100 ms (right) activations. The
insets show corresponding thermal images of a single LED activation taken at
the maximum temperature with 3.2 V input voltage and activation durations of
10 ms and 100 ms, respectively........................................................................ 37

Figure 3-7.

Soak testing result for 25 days: (a) Impedance changes in failed and successful
electrodes during the 25 days of soak testing. The impedance of the failed
electrode increased significantly at day 9. Optical images of (b) successful and
(c) failed electrodes are shown. 3 out of 30 electrodes were failed during the
soak testing. The main cause of the failure was delamination due to the
insufficient protective sealing of Parylene on the periphery of metal
electrodes........................................................................................................... 39

Figure 3-8.

In vivo experiments: (a) mCherry fluorescence in a coronal slice of the
transfected rat V1. In vivo recording set-ups for (b) optical neural modulation
and electrical recording of the Opto-μECoG array and (c) the light-evoked
extracellular recordings using the epidural LED array..................................... 40

Figure 3-9.

In vivo neural recording analysis results: (a) the spectrogram (50-180 Hz) of the
optically evoked ECoG signal recorded through the selected channel (Ch 6)
where the activated LED was located as shown in (b). (c) A series of a matrix
of the average gamma band energy with 50 ms interval. (d) Light-evoked
extracellular recordings at different depths through the carbon fiber
electrode............................................................................................................ 42

Figure 4-1.

Conceptual illustration of the multi-LED array with 3-D polymer waveguides
arrays for deep brain optical stimulation........................................................... 46

Figure 4-2.

Principle of the droplet backside exposure (DBE) method............................... 47

Figure 4-3.

Fabrication process of (a) multi-LEDs array, (b) 3-D μ-waveguides array, and
(c) slanted μ-waveguides array.......................................................................... 50

Figure 4-4.

Fabricated prototypes of (a) a μ-waveguides array and (b) the assembled array
with SU-8 waveguide. (c) SEM images of the SU-8 microneedle array
fabricated on PDMS substrate. (d) before and (e) after assembly with a slanted
microneedle array with the flexible LEDs array. (f) SEM images of slanted
microneedle structures....................................................................................... 54

xi

Figure 4-5.

Simulation results: (a) irradiance measurements at a 100 μm distance from two
tips with diameters of 10 μm and 100 μm, and (b) normalized total flux
measured at a 100 μm distance from waveguide tips with different tip diameters
(10 – 130 μm).................................................................................................... 55

Figure 4-6.

(a) Experimental setting to study optical throughput at the tip of waveguide, (b)
relative intensity of the blue light (450-495 nm) extracted from the captured
image, (c) the zoom-in view of the boxed area in (b). (d) Experimental setting
to study optical throughput the μ-LED chip without the waveguide, (e) relative
intensity of the blue light (450-495 nm) extracted from the captured image, (f)
the zoom-in view of the boxed area in (e)......................................................... 56

Figure 4-7.

Penetration profile of (a-d) the 3-D μ-waveguides array and (e) the slanted μwaveguides array sin different curvatures of gelatin surface............................ 58

Figure 4-8.

Evaluation of mechanical robustness of microneedle arrays using the axial
loading test: (a) from the top left, force measurement vs. displacement of the
microneedle array with different base diameters of 250, 200, 100, and 150 µm.
(b) Needle failure force vs. base diameter......................................................... 59

Figure 4-9.

Histological analysis for long-term biocompatibility evaluation: (a) 4 × 4
slanted microneedle array was implanted on rat V1. (b) A coronal section (5 µm
-thick and stained with H&E) of the V1 after 6 months implant...................... 61

Figure 5-1.

Principle of the proposed droplet backside exposure (DBE) method............... 65

Figure 5-2.

An analytical model for designing slanted microneedle arrays, approximated
based on the geometry of an ideal partial sphere droplet: (a) a diagram of an
ideal partial sphere, or a spherical cap; (b) the ideal droplets, formed on the
hydrophilic surface, with a minimum CA (αmin) and a maximum CA (αmax),
respectively; (c) the relationship between two partial spheres, V1 and V2 ....... 68

Figure 5-3.

Finite element stimulation result of gravity effect on the maximum height of
SU-8 droplets with different base diameters..................................................... 71

Figure 5-4.

The core fabrication process of the DBE method and its variation: (a) the core
DBE fabrication method; (b) a slanted hollow microneedle fabrication process
flow.................................................................................................................... 74

Figure 5-5.

(a) Dispensed SU-8 droplets on PDMS. (b) Example of CA measurement. SEM
images of (c) the inclining profile of the microneedles with a 35.2° CA, and (d)
the aerial view of a circular microneedle array fabricated using a 20 μL droplet.
(e) and (f) SEM images of a hollow microneedle array.................................... 76

Figure 5-6.

Dimension control of the DBE method: (a) comparison between the estimated

xii

heights and the measured heights of the fabricated microneedles with the
standard deviations of less than 5%. (b) Estimated and measured heights of
microneedles formed at different distances from the center of the droplet: 780
μm, 1440 μm, 2130 μm, and 2830 µm. (c) Dependency of base (left) and tip
(right) size on SU-8 droplet volume. This experiment tested microneedles
fabricated using four aperture diameters of 140, 120, 100, and 80 μm, at 780,
1140, 2130, and 2830 μm distance from the center of the droplet, respectively,
with 24 different volumes ranging from 10 µL to 27 µL.................................. 79
Figure 6-1.

Proposed behavioral animal experiment to test artificial visual perception using
cortically optical neural stimulation of the rat V1............................................. 85

Figure 6-2.

In vivo tests with (a) a natural visual stimulation and (b) a cortically optical
stimulation......................................................................................................... 89

Figure 6-3.

Instantaneous phase measure across trials based on Hilbert transform of the
ECoG signals bandpass filtered at 4 – 8 Hz: (a) visual I (106 trials at channel 4),
(b) Visual II (114 trials at channel 4), and (c) Opto I (38 trials at channel 10).
The black line on (a), (b), and (c) indicates stimulation ON. (d) Instantaneous
phase and their corresponding color codings are shown................................... 90

Figure 6-4.

Broadband (1 – 150 Hz) phase-locked synchrony measured by RID-TFPS: a
black line indicates stimulation ON. Phase-locked synchrony measures from (a)
Visual I, (b) Visual II, and (c) Opto I................................................................ 92

Figure 7-1.

Concept diagram of wireless optrode array with integrated μ-LEDs................ 96

Figure 7-2.

(a) Simplified block diagram of a conventional inductively-powered device
combined with an array of LEDs for wireless optogenetics and (b) simplified
block diagram of the wireless switched-capacitor stimulating (SCS) system to
efficiency drive the μ-LED array...................................................................... 97

Figure 7-3.

3-D model for in vivo optogenetics experiments with the SCS system. Inset:
optrode array with micro-LEDs for optical stimulation and penetrating
electrodes wrapped around the waveguides for neural recording................... 100

Figure 7-4.

(a) Fabrication process flow for making the slanted optrode array with
integrated μ-LEDs and (b) the profile of the optrode...................................... 101

Figure 7-5.

Fabricated prototypes of (a) SEM images of a slanted microneedle structure, (b)
32-channel optrode array, (c) microscopic image of individual optrode, and (d)
optrode array coupled with μ-LEDs................................................................ 103

Figure 7-6.

Schematic of in vivo animal experiment set-up with the wireless SCS
system.............................................................................................................. 104

xiii

Figure 7-7.

SCS system GUI in LabVIEW to wirelessly control the SCS parameters. Inset:
LED driving voltage waveform and its controllable parameters.................... 105

Figure 7-8.

Light-induced LEP recording driven by 2.7 Vpeak and 3.2 Vpeak for 100 ms: (a)
LED driving voltage, VLED, for in vivo optogenetics with SCS (b) light-induced
local field potentials (LFP) with VLED = 2.7 Vpeak and 3.2 Vpeak, and (c)
instantaneous phase in low frequency band (1-25Hz), and (d) their
corresponding color-coding............................................................................. 107

Figure 7-9.

Fabrication steps of a hermetically-sealed chronic implant on a flexible
polyimide circuit............................................................................................. 109

Figure 7-10.

(a) Discrete SCS chronic implant and (b) accelerated soak testing in saline with
wireless power and data transfer..................................................................... 111

Figure 8-1.

Basic design of battery-powered wireless optical neuromodulator................ 115

Figure 8-2.

Design of the magnetic-switch based controller and a schematic of the LED
driver circuit.................................................................................................... 116

Figure 8-3.

Design of the IR-switch controller and the schematic of the LED driver
circuit............................................................................................................... 117

Figure 8-4.

Design of IR-MCU based controller and the schematic of the LED driver
circuit............................................................................................................... 118

Figure 8-5.

Summary of GPIAS for measuring tinnitus (adapted from [172]): (a) an animal
is startled in the presence of a background noise plus a startle pulse. (b) A silent
gap 100 ms before the startle pulse can reliably inhibit the reflex. (c) Due to
tinnitus, the gap is not easily detected............................................................. 119

Figure 8-6.

Optical characterizations of the battery-powered wireless optical
neuromodulator............................................................................................... 122

Figure 8-7.

Chronic implantation of the LED controllers in freely-moving mice............. 122

xiv

CHAPTER 1
A hybrid neural interface with integrated LEDs for
optogenetics
Brain machine interfaces (BMIs) provide a direct communication pathway between the brain and
external devices, and enable a promising therapy to restore motor control in severely disabled
patients suffering from events such as spinal cord injury, stroke, and traumatic major
amputations [1]. Current BMIs techniques primarily use visual feedback to drive prosthetic
devices, and lack direct sensory feedback from artificial limbs into the brain [2]-[5]. With these
single-modality interfaces, experimental and clinical results have demonstrated the ability of
non-human primates and severely disabled patients to use brain activity alone to control a robotic
arm to reach for and grasp objects [6], [7]. However, dramatic performance degradation, caused
by a lack of proprioceptive and somatosensory feedback, has been observed in patients suffering
from complete large-diameter-fiber neuropathy, who experienced difficulty executing
sophisticated tasks (e.g. drinking from a cup) [8]. Therefore, a reliable and bi-directional neural
interface is needed to close the BMIs control loop and improve recognition and control abilities
of the prosthetic devices [9], [10]. To date, significant progress has been made in developing bidirectional neural interfaces with both recording and stimulation modalities for closed-loop
BMIs.
Existing artificial neural interfaces mainly use electrical signals to evoke sensation in the
central- and peripheral- nervous systems (CNS and PNS respectively). These electrical systems
have become powerful tools in electrophysiology and clinical neuroscience to treat neurological
events, injuries, and diseases. For example, applications include upper and lower limb prostheses

1

for spinal cord injury and stroke; bladder prostheses; cochlear and brain-stem auditory
prostheses; retinal and cortical visual prostheses; cortical recording for cognitive control of
assistive devices; vagus nerve stimulation (VNS) for epilepsy, depression, and tinnitus; and deep
brain stimulation (DBS) for essential tremor, Parkinson’s disease, epilepsy, dystonia, and
depression [11]-[13].

Figure 1-1. Types of neurophysiological signals and their temporal and spatial resolutions
adapted from [14].
Depending on the location of the electrodes, the neurophysiological signal can be
classified into three categories: electroencephalogram (EEG), electrocorticogram (ECoG), and
extracellular recordings as shown in Fig. 1-1. Specifically, EEGs recorded from the scalp are the
summation of the synchronous activity of large populations of cortical neurons. EEG recordings
provide easy and non-invasive access to cortical neural activity, thus have been used extensively
in various fields, such as neuroscience, cognitive science, clinical applications, and BMIs.
However, the EEG recording suffers from poor spatial and temporal resolution caused by the
filtering effect of the skull as well as artifacts from muscle movements [1].
Extracellular recordings—the observation of action potentials generated by a single or an
ensemble of neurons and local field potential (LFP) from all nearby dendritic synaptic activity—
can be obtained by direct implantation of voltage-sensing micro-scale electrodes. Action
2

potentials from a single neuron or a population of neurons have been used in many applications,
and are considered the gold standard in electrophysiological studies [15]-[17]. Despite its many
benefits, the extracellular recording requires invasive surgery to implant the recording electrodes,
which surgery can result in complications and permanent tissue damage.
Similarly to EEG, ECoG also is the representation of synchronous cortical neural activity,
but the neural activity is directly recorded from the cortical surface to circumvent the rapid signal
attenuation effect of the skull [1]. ECoG has higher spatial resolution, broader bandwidth, and
higher amplitude than does EEG [18]. Recently reported are micro-ECoG (μECoG) BMIs
devices enabling a vey high spatial resolution (360 channels in 2 × 3 cm) of neural recording
[19]. Further, the ECoG is free of artifacts caused by muscle and eye movements consistently
present in EEGs of awake and moving subjects. Finally, epidural ECoG recordings are less
invasive than extracellular recordings, reducing risks of stroke, hemorrhage, and infection
associated with penetrating electrodes, as well as signal instability in long-term use. A bidirectional ECoG interface also has been reported, showing that electrical pulses delivered
through a μECoG electrode effectively can evoke sensation on the primary somatosensory cortex
(S1) [20]. However, electrical stimulation through micro-scale electrodes often requires high
current to activate neural activity, which can cause toxic redox reactions and damage to both the
tissue and the electrodes. Moreover, the inherent limitations of electrical stimulation such as
unpredictable current pathway, electrical artifact, non-specificity, and degradation of electrodes
increase demands for a new technology [21].
Emerging optogenetics technology uses direct light stimulation to activate or inhibit
genetically modified neurons. Optical stimulation provides many distinct advantages including

3

cell-type specificity, sub-millisecond temporal precision, rapid reversibility, elimination of
electrical artifacts, and prolonged lifetime due to hermetic sealing of light sources [21], [22].

Figure 1-2. Different optogenetics-based hybrid neural interfaces for optical stimulation and
electrical recordings of cortical neurons: (a) a conceptual illustration of the optical
neuromodulator integrated with LEDs (b) Opto-μECoG array (c) 3-D multi-LED arrays coupled
with micro optical waveguides (d) and slanted waveguides.
Current optogenetics-based stimulation interfaces mainly use laser- or light emitting
diode (LED)-coupled optical fiber [23]-[25], micro-LED arrays [26], and a laser beam focused
through a microscope [27] as their light sources. For experiments with freely behaving subjects,
however, only limited light delivery methods such as a laser-coupled optical fiber and a headmountable single LED system [28] are available. These systems’ poor spatial resolution limits
their functionality, and the tethered optical fiber greatly restricts subjects’ natural behaviors.

4

To address these limitations, a series of hybrid neural interfaces based on a polymerbased flexible micro LEDs (μ-LEDs) array has been developed that would provide a unique
solution to the current demand for multichannel, bi-directional neural interface devices.
Depending on interface location, there are three different neural interfaces: Opto-μECoG
array for epidural stimulation, three dimension (3-D) waveguide array for deep cortical
stimulation, and slanted 3-D waveguide array for multi cortical layers stimulation. The OptoμECoG array contains a transparent μECoG electrode array with integrated bare chip μ-LEDs
enabling optical stimulation and electrical recording. This author has demonstrated that a 32
individually addressable μ-LED array can evoke neural activities not only within the superficial
cortical layer, but also within deep cortical layers (~ 600 μm deep) [29]. Although this 32
channel μ-LED array significantly improved spatial resolution of the optical stimulation, its
depth resolution had a limitation.
A 3-D waveguide array integrating µ-LED chips with microneedle-shaped waveguides
was developed to minimize scattering of LED lights in the brain tissue and achieve high spatial
resolution in 3-D [30]. This author further improved the spatial resolution of the 3-D waveguide
array in depth by a varying-length waveguide array [31]. Integration of individually addressable
µ-LEDs with slanted microneedles allows precise light delivery to the target neurons in
individual cortical layers. To create out-of-plane microneedles with various lengths, a polymerbased fabrication method was developed, utilizing the height variance in a dome-shaped, epoxybased, negative photoresist (SU-8) structure. This method, dubbed a droplet backside exposure
(DBE), independently can control the length, and tip/bottom diameters of individual
microneedles without using a specialized machine or complex fabrication techniques. Fig. 1-2

5

shows a comparison of three different hybrid neural interfaces. Since these arrays are based on
integrated light sources, they can be wirelessly powered and controlled.

Figure 1-3. The overall design of the wireless optrode neural interface.
To demonstrate the wireless capability of the arrays, this author and colleagues proposed
a wireless-powered, multichannel optrodes array that integrated µ-LEDs with microneedle

6

waveguides [32]. The waveguide arrays can be converted into optrodes, opto-electrodes for
combined electrophysiological recording and optical stimulation array by utilizing metallic
shielding layers as a recording electrode to achieve bi-directional hybrid neural interfaces.
Recording electrodes wrapped around the waveguide cores permit simultaneous measurement of
neural responses from various depths, upon optical excitation. Arrays were inductively powered
and controlled by the wireless switched-capacitor stimulator (SCS), designed to improve power
efficiency [33]. Fig. 1-3a and b depict the concept and design of the slanted optrodes array.
Based on the development of hybrid neural interfaces, this author is designing an
optogenetics-based cortical visual prosthesis for treating all forms of blindness. With our hybrid
neural interfaces, the core hypothesis that epidural optical stimulation of microbial opsin
expressing neurons in the primary visual cortex (V1) can induce phosphine, artificial visual
sensation, and perception, will be tested. In preliminary work, capability of the Opto-ECoG array
for simultaneous optical stimulation and epidural recording of neural activities of the rat V1 has
been demonstrated [34]. Several in vivo animal experiments have been conducted to study the
correlation between neural activities in the V1 and light-evoked phosphenes including stimulusdriven neural data analysis, a chronic device development for freely behaving animal experiment,
and a behavioral test in response to external stimuli.
This dissertation has organized the details of development of these neural interfaces and
simulation and/or in vivo animal experiment results that have been demonstrated to evaluate the
feasibility and functionality of the proposed hybrid interfaces for cortical visual prosthesis. The
remainder of this report is structured as follows. In Chapter 2, background of optogenetics,
available light delivery mechanisms, and specific examples about therapeutic treatment of
neurological diseases based on optogenetics are reviewed. Chapter 3 presents design

7

specifications, device fabrication, and characterization of the Opto-μECoG array. In vivo
experiments were performed to investigate the capability and reliability of the Opto-μECoG
array. In Chapter 4, the design objectives, device fabrication and characterization of the 3-D
waveguide arrays are discussed. A new fabrication process, called droplet backside exposure
(DBE) method, for slanted microneedle structure, is presented in Chapter 5. Based on the devices
and technologies discussed in previous chapters, a detailed plan for development of cortical
visual prosthesis and current efforts in light-evoked neural data analysis are presented in Chapter
6. Chapter 7, as part of the development of cortical visual prosthesis, demonstrates chronicallyimplantable wireless flexible optrode array with multichannel μ-LEDs for selective optical
stimulation of cortical neurons and simultaneous recording of light-evoked neural activity.
Finally, in Chapter 8, the dissertation is closed via reference to a recent progress of the
development of cortical visual prosthesis including the design of a behavioral animal experiment
and battery-powered wireless neurostimulators for the experiment.

8

CHAPTER 2
A review of optogenetics and light delivery methods
2.1 Introduction
Optogenetics is a new neuronal modulation method allowing direct control of the activity of
neurons by light. This can be achieved by genetically modifying the cell-types of interest to
produce specific opsins, light-sensitive proteins. Once targeting cells express these opsins, they
become light sensitive, and direct light exposure to the cells trigger neural activity [35]. Opsins
are genetically encoded into the cells of interest, and this method allows targeting the specific
cell population. Because each opsin has different characteristics, depending on the expressed
opsins the cell can be stimulated or inhibited with light exposure of specific wavelengths [22].
Optical stimulation based on optogenetics is considered more beneficial than conventional
electrical neuronal stimulation, because it permits activation or inhibition of specific types of
neurons with sub-millisecond temporal precision, eliminates electrical artifacts, and potentially
has a prolonged lifetime due to hermetic sealing of light sources.
This chapter reviews some recent advances in optogenetics and related technologies
including the progress in development of new tools of optogenetics via genetic engineering,
development of light delivery methods, and applications of optogenetics in the study and
treatment of neurological diseases.

2.2 Light sensitive proteins
Optogenetics utilizes microbial opsin to modulate excitable cells. These opsins shown in Fig. 2-1
have been isolated from various microorganisms and plants, and channelrhodopsin (ChR) and

9

halorhodopsin (HR), naturally occurring photosensitive proteins, at first were used to induce
activation or inhibition of neural activity in mammalian neurons [22], [36]. Since then, protein
engineering has generated an expanded optogenetics toolbox for more precise and effective
neuronal modulation including single-component proteins activated by various wavelengths of
light, with various ion conductance regulation properties, enabling broad experimental
configurations and opportunities [21], [37], [38]. Available single-component optogenetics tools
can be categorized into four groups: excitation, inhibition, bi-stable modulation, and control of
intracellular biochemical signaling in neurons and other cell types.

Figure 2-1 Basic properties of single-component optogenetics tool adapted from [39].
2.2.1 Excitation
ChRs are light-gated, non-specific cation channels in unicellular green algae that allow elective
depolarization of genetically targeted cells [22], [40]-[45]. Molecular engineering of ChR has
been very active over the last few years, and various ChR variants with distinct spectral and
kinetic properties have been developed. Particularly, two mutations, H134R and T159C,
achieved increased photocurrent amplitude by a factor of 1.5 to 2 in pyramidal cells [46], [47].
However, this improvement in photocurrent amplitude resulted in slower off-kinetics of 30 – 60

10

ms in neurons [48]. One of the key amino acids controlling the protonation state of the retinal
chromophore is a glutamate at position 123 in the channelrhodopsin-2 (ChR2) sequence, and
substitution of an alanine or threonine at this position generated E123A/T mutants, in which the
photocycle is accelerated to an apparent off-kinetics of 4 – 6 ms [42]. These mutations can
increase the reliability of high frequency evoked action potentials.
Table 1. Known single-component optogenetics tools for excitatory

Initially available ChR variances were activated with limited light spectra that typically
peak at 450 – 490 nm [43], [49]. These blue wavelengths have limited penetration depths into

11

neural tissue in mammalian systems, as they are strongly absorbed and scattered more strongly
than yellow-red wavelengths [50]. The practical issue of efficient light delivery can be solved by
creating a red-shifted ChR with spectral peaks near or above 600 nm, where light absorption and
scattering by the tissue drop off steeply [35], [51]. An opsin from Volvox carteri (VChR1),
which shares homology with ChR2 and works as a cation channel, can be activated by redshifted light [51]. However, its relatively small currents due to low expression in mammalian
neurons have limited in vivo adaptation. A new family of engineered chimeric opsin variants
(C1V1) composed of channelrhodopsin-1 (ChR1) and VChR1 fragments implements fast and
potent optical excitation at red-shifted (535 – 545 nm) wavelengths [52]. Red-activatable
channelrhodopsin (ReaChR), a new variant of ChR, has improved membrane trafficking and
expression in mammalian cells, a more robust spectral response above 600 nm and an enhanced
steady-state response to light with wavelengths longer than 600 nm [53]. Since these ChR
derivatives can be activated without affecting ChR2, there is a possibility of combinatorial
excitation in vivo [52].
2.2.2 Inhibition
In principle, multiple different cell populations could be involved in the same circuit or
behavioral property. For this reason, neural inhibition, in addition to these tools for neural
excitation, can be used for studying the targeted cell population. NpHR is a light–activated
chloride pump, hyperpolarizing the targeted neuron upon activation, with photocurrent peak at
590 nm and it can be found in multiple species of halobacteria [23]. Thus NpHR and ChR
together create a powerful toolbox for neuroscientists by enabling multiple-color optical control
for activating, silencing, and desynchronizing neural activity. Although optogenetics inhibition
with NpHR was shown to operate well in freely-moving worms and in mammalian brain slices

12

as well as cultured neurons [23], [54], it suffered from membrane trafficking problems that must
be addressed for use in behavioral studies in intact mammals.
Table 2. Known single-component optogenetics tools for inhibitory

Optimization of membrane targeting was required mainly for the inhibitory microbial
rhodopsins, because most were derived from prokaryotic species in which membrane proteins
are produced and transported using mechanisms significantly different from those in mammalian
cells. This caused a tendency to accumulate in cells’ Endoplasmic reticulum (ER). To improve
proper assembly of these rhodopsins, Golgi and ER export motifs have been added to each end
of the NpHR sequence [55]. That addition of the export motifs from mammalian ion channels
reduced protein aggregation and increased the number of rhodopsin molecules reaching the
plasma membrane [56]. Consequently, the most commonly used tools for neural silencing are the
two enhanced versions: NpHR3.0 and Arch3.0 [37], [56], [57]. In a recent study, Chuang et al.
introduced a red-shifted cruxhalorhodopsin, Jaws, for noninvasively mediating transcranial
optical inhibition [58]. It is derived from Haloarcula (Halobacterium) salinarum (strain Shark)
and engineered to result in red light–induced photocurrents three times those of earlier silencers.

13

Table 3. Known single-component optogenetics tools for bi-stable excitatory

2.2.3 Bi-stable
Minimum activation light intensity for ChR2 mutations designed for faster off-kinetics was
increased from 1 mW/mm^2 to 5 mW/mm^2, resulting in potential tissue over-heating [59]. To
address this heating issue and to increase effective light sensitivity of optogenetic excitation,
light-switchable variants have been engineered. Step-function opsins (SFOs) are a family of ChR
mutants displaying bi-stable behavior: prolonged activity after termination of the light stimulus
by bringing membrane potential close to the action-potential threshold and increasing the
probability of spiking [60]. This can be achieved by mutations at positions C128 and D156.
Single mutations C128S and D156A slow the apparent off-kinetics of ChR2 to 120 and 250 s,
respectively [60], [61]. Combining both mutations generated a variant with a closing time
constant of 30 min, and this stabilized SFO (SSFO) induces peak currents of >200 pA [52]. The
SFOs also can be deactivated by a pulse of yellow light [60], hence these mutants can be
activated by a short blue light pulse, then can be deactivated with a yellow light pulse allowing
for precise control of network dynamics. Another advantage of SFOs is light sensitivity of the

14

expressed cells, due to photon integration during the entire duration of blue light illumination.
This property renders the SFOs potential minimally invasive tools for stimulating large brain
regions and deep brain structures.
2.2.4 Signaling
Table 4. Known single-component optogenetics tools for bio signaling

Following the success of applying microbial rhodopsins as optogenetics tools in mammalian
neurons, similar approaches have been applied to engineer a range of light-driven actuators
allowing manipulation of diverse cellular processes [62]. One common approach involves
addition or substitution of enzyme domains with photoreceptor units, which exhibit structural
changes with optical stimulation [63]. In Drosophila, bPac, photo-sensitive adenylyl cyclase,
expression increases the level of the signaling molecule cAMP by a factor of ten on blue light
illumination [64]. Recently, the toolbox was extended by a light-inducible transcriptional
effector for endogenous gene expression. A light-inducible dimerization of the cryptochrome
Cry2 with its native truncated binding partner CIBN [65], was fused to a transcription effector

15

domain and a DNA recognition domain, respectively. Upon blue light illumination, the DNA
bound CIBN recruits the Cry2-fused activator domain, and thereby can increase the transcription
of target genes [66]. These and other tools have extended the range of optogenetics manipulation
beyond the simple control of neuronal excitability and are likely to contribute significantly to a
range of experimental applications [64].

2.3 Light delivery methods
Optogenetics has created a demand for optical devices targeting delivery of light not only on the
surface but also to subregions of the brain. For surface light stimulation, several wide-field
delivery systems have been proposed such as steering a focused laser beam using acousto-optic
deflectors [67], digital micro-mirror devices (DMDs) [68], [69], and computer-generated
holograms (CGHs) [70]. To deliver light inside the tissue, laser- or LED-coupled optical fibers
are implanted in the brain [71]-[73]. Recently, more advanced methods based on
microfabrication technology including channel waveguide arrays, arrayed optical fibers [74],
[75], multiwaveguide fabrication on a single substrate [24], and polymer-based flexible microLEDs array [76] have been developed to increase spatial resolution of light in the threedimensional structure of the brain. These methods can be categorized into three groups:
waveguide-based systems, light distribution systems, and optrode-based systems, and details are
discussed.
2.3.1 Waveguide-based systems
Since the most convenient and affordable lightguides for transmitting light over long distances
are optical fibers, they are the most popular waveguide-based system for deep brain stimulation.
In most applications, the beam of a solid-state laser or super luminescence LEDs is coupled to a
multimode step index fiber, which guides and delivers the optical power to the region of interest.

16

Diode lasers and diode-pumped solid state (DPSS) lasers are in common use in optogenetics
studies [22], [72], [77]. Main advantage of lasers is their highly collimated beam, which can be
focused to a very small focal point. This is important for efficient coupling to thin optical fibers
with core diameters of 50 microns or less. In addition, the narrow spectral width is helpful for
selectively illuminating multiple opsins.

(a)

(b)
(c)
Figure 2-2. Waveguide-based light delivery methods: (a) laser- or LED-coupled optical fibers
and the light delivery through an implanted cannula [22], (b) implantable three-dimensional sets
of silicon oxynitride waveguides [79], and (c) multipoint-emitting optical fibers [81].

17

Using a LED illumination system for optogenetics studies provides many advantages
including illumination stability and fast light switching [78]. Because LED systems do not
require a resonant cavity like lasers, they do not have the noise related to emission modes or
instability from back reflections that can be found with diode lasers. Precise temporal activation
of LED systems is another advantage as a light source for optical stimulation, since optogenetics
studies require well-defined temporal control. Most DPSS laser systems with TTL modulation
input become unstable at fast switching rates, so mechanical fast shutters are necessary. LED can
be directly modulated with TTL input for fast switching with a rise/fall time of microseconds.
Electronically driven LEDs are particularly suitable for integrated systems such as prosthetic
devices.
Optical fibers can be guided into the brain through an implanted cannula as shown in Fig.
2-2a [22]. In this method, viral infection and light delivery are in range, since the guide cannula
is also used for virus injection. Light also can be delivered through implanted optical fibers [71].
In this case, a small piece of optical fiber is permanently implanted into the brain over the region
of interest, and light is delivered into this fiber through an optical fiber connector mounted onto
the animal's head. When compared with the cannulated light delivery, this approach has dual
benefits: reducing the implant diameter, and coupling and de-coupling of the implanted optical
fiber to a light source done easily with freely-behaving animals.
These methods are simple and reliable, but spatial control of light distribution is limited. It
is possible to make an array of channel waveguides where each waveguide in the array
terminates at a different depth. Recently, implantable three-dimensional sets of silicon oxynitride
waveguides have been developed; raising the possibility of generating 3D distributed light
patterns in the brain [79]. Individual waveguides can be addressed by a matrix of micro mirrors

18

or separately coupled to different light sources, allowing optical stimulation at each point with
tunable wavelength and intensity (Fig. 2-2b). Another approach for 3D light distribution is
physically to move an optical fiber inside the tissue [80]. In this design, a micro actuator moves a
side-opened fiber, the optical fiber with an angled tip resulting orthogonal radiation pattern to the
optical axis of the fiber, inside a glass cannula implanted in the brain prior to the experiment.
This system can deliver different wavelengths at numerous positions inside the tissue. Recently
multipoint-emitting optical fibers have been developed allowing selective and dynamic
illumination of different brain regions along the taper (Fig. 2-2c) [81]. Site selection is achieved
by a simple coupling strategy at the fiber input, and the use of a single tapered waveguide
minimizes implant invasiveness. Light emission is permitted at selected sites along the taper by
locally removing the coating to create “windows.” Illumination with a well-defined modal set at
the fiber input then addresses emission to specific windows along the fiber [81].
While these methods allow spatially selective illumination, they require a complex
fabrication process and/or coupling strategy at the distal end of the waveguides. Moreover,
despite the wide range of proposed devices, only a few have been tested in vivo [75].
2.3.2 Light distribution systems
Distribution of light on the surface can be controlled by spatial light modulators (SLMs). Two
types of SLMs are used to produce distributed patterns of light in the brain: digital micro-mirror
devices (DMDs) based SLMs and liquid crystal display (LCD) SLMs. The DMD system is a
light processor consisting of a large array of micro mirrors. When integrated into appropriate
optical stimulation equipment, each mirror defines a pixel in a binary image that is projected
onto the specimen. Individual mirrors can be switched independently with sub-millisecond
precision [68], [82].

19

A LCD video projector has been modified to be used with lasers and coherent light sources
[83]. In this design, a laser source produces a collimated beam of light, which is phase modulated
by the LCDs and the modulated beam in launched into the optical path of a fluorescence
microscope. The LCD SLM generates a dynamic mask, which reshapes a laser beam to control
the distribution of light on the sample.
Steering a focused laser beam using acousto-optic deflectors (AOD) is other option for a
surface-targeting light delivery method. Basic purpose of an AOD is to change rapidly the laser
beam direction using the acousto-optic effect to diffract and shift the light frequency using sound
waves, and it provides faster temporal resolution, typically on the order of a few microseconds,
which is much faster than typical inertia-limited physically moving mirror-based scanning
instruments [67]. Computer-generated holograms (CGHs) are optical devices enabling the
projection of arbitrary patterns with high power efficiency [70], [84], [85].
2.3.3 Optrodes
An optrode is a novel dual-modality hybrid device consisting of optical fibers integrated with
recording electrodes [86]. These devices are employed primarily for recording single-unit and
multi-unit activity in anesthetized animals [74], [87], [88]. To introduce a fiber into a Utah-style
multi electrode array (MEA), one of the electrode positions can be replaced with an opening
(Fig. 2-3b) [86]. These devices often employ tapered fibers, which allow for reduction of the
overall device dimensions but significantly increase optical losses and reduce the numerical
aperture leading to reduced illuminated tissue volume. Silicon multi-electrode probes can be
integrated with optical fibers by simply attaching the fibers on the electrode shank with an
adhesive, following the fabrication process (Fig. 2-3 a) [74], [88]. These dual modal interfaces
also can be extended by combining a microfluidic channel for fluid delivery [89]. A tungsten

20

microelectrode-based optrode that encloses optical fibers within its insulation glass was
developed for targeting deep brain structure (Fig. 2-3d) [90]. Optical fibers and a tungsten wire
were tightly bound to each other and integrally coated with a smooth, thin layer of glass. This
design satisfied the structural requirements for use in deep brain structures.
These optrodes require an external light source, and a connector design remains one of the
main challenges for simultaneous neural recording and optical stimulation, due to additional
optical fiber coupling.

(a)

(b)

(c)
(d)
Figure 2-3. Optrode-based light delivery methods: (a) multi-array silicon probes with integrated
optical fiber [74], (b) a Utah-style multi-array with an optrode [86], (c) a neural probe with
monolithically integrated electrodes and waveguide [88], and a tungsten microelectrode-based
optrode that encloses optical fibers within its insulation glass [90].

21

(a)

(b)

(c)
Figure 2-4. Polymer-based light delivery methods: (a) the chip level integration technology [96],
(b) a polymer-based single optrode shaft integrated with microfluidic channel [89], and (c) a
flexible optrode with integrated micro-LEDs using micro-contact printing process [99].
2.3.4 Polymer-based systems
Since the majority of optogenetics neural probes integrate rigid silica fibers, their mechanical

22

reliability would be compromised by optoelectronic integration. In addition, recent interest in the
peripheral nervous system as a potential early therapeutic target for optogenetics demands more
flexible and biocompatible platforms for the advancement of combined long-term optogenetics
and electrophysiological studies.
Over the past several years, a number of technologies have developed to address these
needs. Silicone resins (Polydimethylsiloxane (PDMS)) have been employed successfully as
substrates for metal electrode arrays [91], [92]. This has allowed development of chronically
implanted peripheral nerve cuffs capable of single-unit recordings from the nerve surface.
Similarly, Parylene-C has been employed as a flexible substrate and as biocompatible packaging
materials in various neural interfaces [93]-[97]. Li et al., developed an embedded chip
technology to integrate commercially available IC chips with polymer substrates and passive
components [95]. A placement cavity is etched on the Parylene-on-silicon substrate for chip
placement. Once the chip is fixed inside the cavity, metal is deposited and a photolithography
step is done to electrically connect pads on the chip to outside components (Fig 2-4a). The
Parylene substrate to be integrated with the chip is a Parylene-metal-Parylene sandwich layer,
that has been well characterized both in vitro and in vivo and is also approved by the Federal
Drug Administration (FDA) for long-term human implantations. Several chronic and acute
devices tested with histology results have been published to verify overall biocompatibility and
feasibility of the sandwich layer technology.
The micro-contact printing process developed by Rogers and colleagues has been
employed recently in fabrication of sophisticated integrated structures on highly flexible silicone,
polyimide, and silk fibroin substrates [76], [98]. This approach takes advantage of the mature
chip-scale semiconductor processing and combines it with patterned interconnects which enable

23

release and transfer of devices fabricated on wafers onto flexible substrates. In a recent study, a
flexible optrode, shown in Fig 2-4c, was thus fabricated, integrating a single recording channel
with LEDs, photodetectors, and a thermal sensor [99].

2.4 Specific studies based on optogenetics and their methods
Traditionally, neurological and psychiatric diseases of the brain were considered as loss or
disruption of function due to changes in cell mass (neurodegenerative diseases), cell function
(memory loss), neurotransmitter systems (Parkinson’s disease), or perturbed connections
(different types of addiction) [100]. Optogenetics has been used to identify clinically-relevant
therapeutic targets to understand normal and disrupted function of neuronal circuits by assessing
whether the activation or silencing of certain neurons treats certain disease states in animal
models. Using microbial opsins and devices optimized for delivery of light into the brain as
described above, researchers can monitor how specific cell types, projection pathways, and
regions contribute to Parkinson’s disease, epilepsy, and other human brain disorders in rodent
models [101], [102].
Optogenetics-based circuit mapping has been initiated to understand local connectivity
amongst various neuronal cell types [103], [104] and long-range connectivity between brain
regions [105]. Activation and inhibition of these connections locally and at a distance may alter
neuromodulation balance.
Understanding cortical circuits requires mapping connections between specific populations
of neurons and dendritic locations where synapses occur. ChRs have been used to explore both
synaptic potentiation and depression [103], [106], providing deeper understanding of circuit
development and pathologies. Specifically, ChR2-assisted method [105] was developed to map
spatial distribution of synaptic inputs, defined by presynaptic ChR2 expression, within the

24

dendritic arborizations of recorded neurons. This study demonstrated high specificity in the
subcellular organization of excitatory circuits in a mouse model.
Distribution of local neuronal connections can affect regional inhibition versus excitation,
and the SSFO was used to study a mechanism of different observations of changes in cortical
excitatory/inhibitory balance [107]. This phenomenon might be linked to physiological states
such as relative loss or hypofunction of medial prefrontal cortex (mPFC) inhibitory interneurons,
altered gamma oscillations, and social and cognitive dysfunction in diseases such as autism and
schizophrenia [107]. Using targeted expression of the SSFO in both excitatory and inhibitory
cells of mPFC, a region that exerts significant top-down control over decision-making, impulse
control, and many social behaviors, the balance of excitation and inhibition in this region could
be altered selectively [35], [108], [109]. These studies have provided support for interneurons’
role in the generation of gamma and in sharpening cortical sensory transmission and enhancing
mutual information in connected cells. These results imply that loss or aberrant expression of
inhibitory cortical cells, often observed in psychiatric disease, may lead to cognitive
derangements seen in schizophrenia and autism [38].
A combination of circuit manipulations using optogenetics and behavioral phenotypes has
been used to determine projection-specific targets within normal function and disease. Other
conventional interventions such as pharmacological, transcranial magnetic stimulation, and
electroconvulsive therapy were unsuccessful, partially due to lack of specificity of the treatment
and understanding of the appropriate target circuit sites for intervention [100]. mPFC not only
involves previously-mentioned social dysfunctional disorders, but also sends multiple projections
to subcortical nuclei associated with depression. Warden et al. optically defined two
heterogeneous circuits alternately leading to impulse control disorders and amotivational states,

25

initiated in mPFC [110].
Combining optogenetics with behavioral read-out methods was used to study effectors of
nuclei within a circuit, and the relationship between patterns of activity within these nuclei and
behavioral phenotype. For example, sleep disorders (e.g., narcolepsy) and consciousness (e.g.,
delayed emergence from anesthesia, coma) arise from derangements in pathways of the arousal
circuit, a series of interconnected subcortical nuclei, and Adamantidis et al. demonstrated the
mechanism using the combination of optogenetics with behavioral read-out method [77]. Direct
stimulation of hypocretin-producing neurons within ChR2 expressed lateral hypothalamus (LH)
increased the probability of sleep-to-wake transitions, and shortened the latency to wakefulness
in a rate-dependent fashion. A similar causal role for norepinephrine in arousal was demonstrated
by using a Cre-driver line to express ChR2 in the locus coeruleus (LC) [111]. These studies
enable new targets for the treatment of narcolepsy, other disorders of sleep and arousal, and
delayed emergence from general anesthesia.
The optogenetics approach also can be combined with functional magnetic resonance
imaging (fMRI) [112]. Optogenetics functional magnetic resonance imaging (ofMRI) [112][115] technology combines optogenetics control with high-field fMRI readout enabling cellularlevel precision in control while whole brain network function can be read out in vivo. While
standard fMRI provided important clues to, and insights into, brain function including mapping
of visual, sensory, motor areas, and association of brain activation changes relevant to
neurological diseases, they were limited in accuracy due to the brain’s densely interconnected
network of heterogeneous cellular populations. ofMRI addresses this problem by combining
genetically-targeted control of neurons with a method capable of whole brain functional readout.

26

CHAPTER 3
Opto-μECoG array
3.1 Motivation
Based on the previous chapter’s review of light delivery methods for optogenetics, this author
has proposed a hybrid opto-electrical system combining ECoG electrical recording and optical
stimulation methodology. This polymer-based flexible micro neural interface would provide a
unique solution to the current demand for multichannel, bi-directional neural interface devices.
Whereas some studies have been reported, combining ECoG and optogenetics for cortical
mapping [116] and neural modulation [117], several design challenges limit the development of
a complete BMIs system for a chronic implant. Specifically, current optogenetics-based ECoG
interfaces mainly use laser- or light emitting diode (LED)-coupled optical fiber [118] and SLM
based systems [100] as their light sources, none of which suit BMIs in freely behaving subjects.
In addition, densely arranged ECoG electrodes, made of opaque metals (e.g. platinum), cover a
large portion of the cortical area, so optical stimulation of the cortical area through these opaque
electrodes remains problematic.
To address these challenges, a novel Opto-μECoG array consisting of a transparent
epidural micro-scale electrodes array for electrical neural recording and a LED array for optical
neural stimulation was proposed (Fig. 3-1). The high-density μECoG electrodes array captures
neural activity of large populations with relatively high temporal and spatial resolution. μ-LEDs,
self-assembled on a flexible polymer substrate, enable optical modulation of specific types of
neurons with sub-millisecond temporal precision, and allow the possibility of achieving a truly

27

untethered system crucial for chronic behavioral experiments. Parylene-C serves as the main
structural and packaging material, because its properties - flexibility, optical transparency,
chemical inertness, and biocompatibility - are favorable for implantable devices. In addition,
epidural placement of the device on the surface of the cortex minimizes risk of brain tissue
damage. In this chapter, design specifications of the proposed Opto-μECoG array are described,
followed by device fabrication. Next, the optical, electrical, and thermal characteristics of the
fabricated device are presented. Finally, the in vivo experiments that were performed to
investigate the capability and reliability of the device are explained.

Figure 3-1. A concept diagram of BMI system with an Opto-μECoG array

28

3.2 Design
The concept of the Opto-μECoG array is illustrated in Fig. 3-2. Specific design considerations
were made to address the above-mentioned limitations of the current optogenetics-based neural
interfaces, as discussed below.

Figure 3-2. Concept illustration of the proposed Opto-μECoG array
3.2.1 A multichannel, bi-directional system
The Opto-μECoG array comprises two sub-arrays, each containing 16 channels (4 × 4) on a 2.5 ×
2.5 mm^2 footprint designed to match the bilateral hemispherical area of the visual cortices in
rats. Each channel was constructed by assembling a surface mounted μ-LED (220 × 270 × 50
μm^3, peak wavelength at 460 nm, Cree® TR2227TM, Cree, NC, USA) on top of the
transparent electrode, to allow simultaneous optical stimulation and electrical recording. This
system permits simultaneous recording and/or stimulation of neural activity in individual or
multiple spots. LED light sources potentially could be powered and controlled through wireless
telemetry links to achieve a truly untethered system.

29

3.2.2 Maximum target cortical area of the optical stimulation
Optically transparent epidural electrodes allowed maximum exposure of the target cortical area
for light stimulation. The transparent electrodes were fabricated using indium tin oxide (ITO), a
transparent conductive oxide (TCO) exhibiting low electrical resistivity and high transmittance
in the visible spectrum [119]. ITO is considered not only the most promising electrode material
for various optoelectronic applications [120], but also excellent biocompatible material, based on
cytotoxicity and protein adsorption tests [121].
3.2.3 Maximum spatial resolution of optical stimulation and electrical recording
The LED array must cover the entire targeted cortical area, whereas the effective area of a single
LED must be minimized to achieve high spatial resolution of the optical stimulation.
Arrangement of the μ-LEDs was estimated based on the scattering property of blue light, which
was investigated by quantifying the optical throughput of the LED in the brain tissue. Efforts
have been made to increase spatial resolution of μECoG arrays, but no consensus has been
reached on optimal size and spacing of the electrodes; such size and separation are major
parameters requiring careful design to achieve high spatial resolution and signal-to-noise ratio
(SNR) of the ECoG recording. In this work, the ECoG electrodes were designed with diameter
200 μm and separation 700 μm, based on a recent study by Slutzky et al. [122] of optimal
spacing of subdural, epidural, and scalp electrodes.
3.2.4. Biocompatibility and minimum invasiveness
The micro-optoelectronic device must be biocompatible and flexible for chronic implantation.
Parylene-C was used as the prime structural and packaging polymer, which has shown superb
biocompatibility (ISO 10993 & USP Class VI), great flexibility, optical transparency, chemical
inertness, and low permeability [97]. Epidural placement of the device on the cortex is less

30

invasive than intracortical penetrating electrodes, thus further reducing risk of brain tissue
damage.

3.3 Fabrication

Figure 3-3. Opto-μECoG array fabrication process: (a) processes of μECoG electrode array
fabrication; (b) PDMS stamp fabrication; and (c) LED array substrate fabrication and integration
with the μECoG electrodes.
To reduce fabrication complexity, the ECoG array and the LED array were fabricated and
calibrated separately. The system then was assembled by polymer bonding of individual
components with SU-8. The detailed fabrication process was divided into three steps, as
described below.

31

3.3.1 Fabrication of the μECoG electrode array
Devices were fabricated using a Parylene-metal-Parylene multilayer technique, as shown in Fig.
3-3. (a-1) A 3-inch glass carrier wafer was cleaned by sonication in acetone, isopropanol (IPA),
and deionized (DI) water (3 mins each). Five (5) μm Parylene-C was deposited on the glass
substrate using a chemical vapor deposition (CVD) system (PDS2010 Parylene Coater, Specialty
Coating System, IN, USA). (a-2) DC sputtering of a 0.1 μm thick ITO layer was performed in a
Kurt Lesker Axiss PVD System (a 2-inch diameter ITO target [10 wt% SnO2], base pressure was
5e-6 Torr). The ITO target was oxidized by pre-sputtering with DC power of 50 W in an Ar- O2
gas mixture (20 sccm of Ar and 1 sccm of O2) for 5 mins. Then the ITO layer was wet-etched
with 0.4 M of oxalic acid using a photoresist (PR) mask.
(a-3) A 0.1 μm Au layer was deposited on the substrate by a thermal evaporator (Edward
Auto306, Edwards, UK) and patterned by wet etching to form interconnection leads, as shown in
Fig. 3-4b. (a-4) After that, another layer of Parylene-C was deposited to seal the entire structure.
(a-5) Electrode sites and contact vias were opened by oxygen plasma etching of the Parylene.
3.3.2 Fabrication of a PDMS stamp
(b-1) A 3-inch glass wafer was cleaned and went through a dehydration bake. A ~ 30 μm SU-8
layer was spun onto the wafer and patterned as the mold for fabricating a PDMS stamp. (b-2)
PDMS (Sylgard 184, Dow Corning, MI, USA) was poured over the SU-8 mold to form the
stamp, which contained cavities matching the shape of the LED chip. (b-3) After curing for 40
mins at 95 °C, the PDMS stamp was peeled off the mold. (b-4) Thirty-two LED dies were
aligned in the cavities of the stamp, metal pads facing outward. The assembly method using the
PDMS stamp was adapted from [123], and the yield of the chip assembly was above 96%.

32

3.3.3 Substrate fabrication and assembly
(c-1) A 3-inch glass carrier wafer was cleaned by sonication in acetone, IPA, and DI water (3
mins each), followed by Parylene-C (5 μm in thickness) deposition. (c-2) A 0.3 μm Cu layer was
deposited in the thermal evaporator and patterned by wet etching to form contact pads and
interconnection leads. A PR mask was patterned selectively to expose only the metal contacts for
LED connections. (c-3) Low melting point (LMP) solder (melting point at ~ 62 °C, 144 ALLOY
Field's Metal, Rotometals, Inc., CA, USA) was applied on the contacts. The substrate was then
rinsed with acetone, IPA, and DI water to remove the PR layer. (c-4) The PDMS stamp with the
embedded LEDs was aligned to match the pre-soldered receiver sites on the substrate. (c-5) The
substrate with the aligned PDMS stamp was heated on a hot plate at 90 °C for 30 secs. The
PDMS stamp was removed carefully after the substrate had cooled to 40 °C, leaving the LEDs
attached to the contact pads of the substrate. The substrate was then submerged into a hot acidic
water bath (90 °C, pH of 2.0) for 1 min. This process permits fine adjustments of LED alignment
and the formation of an electrical connection in a self-assembly manner (Fig. 3-4d and 3-4e). (c6) A ~ 50 μm SU-8 layer was spun on the μECoG array as a bonding adhesive. The assembled
LED array was aligned and bonded to the μECoG array, followed by oxygen plasma etching to
define the contour of the carrier substrate.

3.4 Results and discussions
Fig. 3-4 shows an assembled Opto-μECoG interface, with total thickness ~75 μm. Several
experiments were conducted to characterize the optical, electrical, and thermal properties of the
fabricated devices. In vivo acute animal experiments were performed to demonstrate the ability
of the Opto-μECoG interface to simultaneously modulate and record neural activities in the
visual cortex of a rat.

33

Figure 3-4. A fabricated Opto-μECoG array and light scattering test: (a) ITO electrode array, (b)
the assembled Opto-μECoG array, and (c) zoomed-in image of individual modules. (d) LED
array before hot water reflow (left) and after the reflow (right). (e) Light scattering property of
the µ-LED chip was estimated in brain tissue and (f) a normalized relative light intensity was
estimated. Both color and the height (z-axis) indicate blue light intensity.
3.4.1 Optical properties
In this study, 1000-Å-thick ITO was sputtered on a 10-μm-thick Parylene-C film at room
temperature. Optical transmittance of the ITO-Parylene film was measured over the visible
wavelength range (450 – 750 nm) at 37 °C using a custom-built spectrum analyzer. Average
transmittance throughout the spectrum was ~ 80 %, while transmittance at the wavelength of 470
nm was ~ 55 % (Fig. 3-5a). This result was comparable with the ~ 60 % transmittance measured
using a 10-μm-thick Parylene-C film reported in [124], considering the additional light
absorption of the Parylene-C film.
To optimize the design of the LED array for maximizing the effective stimulation area
while minimizing the number of LED chips, the optical throughput of the LED chip in brain
tissue was studied. The LED array was placed on top of the harvested rat brain tissue, and the

34

images of scattered light from an activated LED were captured (Fig. 3-4e) and analyzed using
MATLAB (MATLAB 2013, MathWorks, MA, USA). Relative light intensity across the brain
tissue was labeled with different colors and plotted in Fig. 3-4f. The LED chip was driven by 2.8
V, resulting in a power consumption of 21 mW, a temperature increase of 0.3 °C, and irradiance
of 10 mW/mm^2. The effective area, where 90% of optical output was delivered, was estimated
to be ~ 1.2 mm in horizontal distance and ~ 1.3 mm in depth. Considering the scattering property
of the LED and the 55% transmittance rate of an ITO-Parylene film at 470 nm, the maximum
radius of effective light excitation was estimated to be approximately 300 – 350 μm.
Consequently, the minimum inter-electrode/LED spacing was designed to be 600 – 700 μm.

Figure 3-5. Optical and electrical properties: (a) visible transmission spectra of ITO film and (b)
I-V characteristic of a μ-LED chip.
3.4.2 Electrical properties
Sheet resistance of the ITO film was measured by a 4-point-probe unit (VEECO FPP-5000, NY,
USA). The pre-oxidation step described in the fabrication process significantly reduced the sheet
resistance of ITO films from 88 Ω/sq to 12 Ω/sq, without compromising their optical
transparency [125]. To demonstrate the benefit of the multilayer metal interconnects, the
electrode-electrolyte interface impedances of electrode arrays with only ITO metalization as well

35

as arrays with both ITO electrode and ITO-Au interconnect metallization was tested.
Measurements were done at 1 kHz using a built-in electrode-impedance-testing circuitry in Intan
evaluation board (RHA2116 and RHA2000-EVAL, Intan Tech. LLC, CA, USA). Arrays with
ITO-Au interconnect showed lower impedances ranging from 1 to 5 kΩ, as compared to the
impedances of arrays with only ITO metalization (7 – 20 kΩ). These results demonstrated that
the ITO-metal multilayer metalization not only prevented breakage of ITO leads due to the
brittleness of ITO [117], but also significantly reduced overall impedance of the electrodes.
Power consumption of the μ-LED at various operating conditions was investigated also,
which provides a useful guideline for optical stimulation in in vivo experiments and future
integration of the LED array with wireless power telemetry. The nonlinear current-voltage
characteristic of a single μ-LED was measured with an impedance analyzer (HP 4191A RF,
Hewlett Packard, CA, USA), as shown in Fig. 3-5b. Since a minimum light-energy density of 1
mW/mm^2 must be delivered onto the target area to induce action potentials, a minimum input
voltage of ~ 2.7 V is required to drive a single μ-LED chip, resulting in a total power
consumption of 3.4 mW.
3.4.3 Thermal properties
Tissue damage caused by heat production of an implanted light source is a major concern, since
the epidurally-implanted LED array will interface directly with brain tissue. Therefore,
understanding thermal properties of the LED chip under various conditions is critical.
In this work, temperature variation of the LED array was investigated using a high
definition infrared camera (Delta Therm. HS1570 and software (DT v.2.19)) with different input
voltages (2.7 – 3.2 V) and activation durations (10, 50, and 100 ms, respectively). Thermal
images were taken in a room (at room temperature of 22.7 °C) where ventilation was strictly

36

controlled to minimize ambient thermal noise, and the array was suspended in the air during the
LED activation. Captured thermal images were calibrated and analyzed using MATLAB; results
are plotted in Fig. 3-6. As expected, both the input voltage and duration of stimulation
dramatically changed the temperature profile of the LED. With the lowest input voltage of 2.7 V,
the heat produced by the LED resulted in the local temperature increase of less than 0.1 °C.

Figure 3-6. Thermal properties: Heat dynamics of the Opto-μECoG array at various input
voltages (2.7 – 3.2 V) for 10 ms (left) and 100 ms (right) activations. The insets show
corresponding thermal images of a single LED activation taken at the maximum temperature
with 3.2 V input voltage and activation durations of 10 ms and 100 ms, respectively.
Based on prior work by Andersen et al. [126], 0.5 °C is a conservative limit on
temporarily-allowable temperature rise in brain tissue adjacent to the device. The test results
suggest that a short pulse of activation (2.7 V amplitude and 10 – 100 ms pulse durations) would
satisfy this safety limit. The heat dissipated within 50 ms and the LED temperature returned to
the baseline point (~ 22.7 °C). As the input voltage and the activation duration increased,

37

significant heat was generated by the LED, resulting in higher local temperature, larger heat
conduction, and longer heat dissipation time. Maximum temperature increase was over 9 °C
when the LED was driven by 3.2 V with the 100 ms activation duration. Considering the tradeoff
between heat production and light-energy density of the LED, the 2.7 V input voltage and 100
ms duration was chosen as the stimulation parameters for in vivo animal experiments.
3.4.4 Soak testing
A passive soak testing also was conducted to study electrode array reliability. During
experiments, the device was immersed in saline at an elevated temperature of 37 ºC, then
removed daily from the oven to monitor device failures through measurement of electrical
function and visual inspection. The device was suspected to fail from soak testing when its
impedance was over 1 MΩ. Then the array was tested in saline where a short pulse with 1 mV
amplitude and 10 ms duration was injected. If the signal-to-noise ratio (SNR) of recording
through the electrode was lower than 1 dB, the electrode was confirmed being failed. Three of 30
electrodes failed after 25 days of soaking. From microscope inspection, all 3 failures were caused
mainly by delamination and/or water traps at the Parylene-metal interface due to insufficient
Parylene sealing on the periphery of the electrodes. The 27 electrodes remained functioning
(impedance: 15 – 110 kΩ) even after 25 days of soak testing (Fig. 3-7).
3.4.5 In vivo animal experiments
In vivo acute animal experiments were conducted to evaluate surgical and functional
applicability of the devices at the cortical interface. All procedures were approved by the
Institutional Animal Care and Use Committee (IACUC) at Michigan State University. Prior to
the acute experiments, rodent subjects (Sprague-Dawley rats: 250 – 400 g) had neurons in their

38

primary visual cortex (V1) transfected with ChR2 to enable light sensitivity. For the viral
transduction, each rat was anesthetized with a mixture of ketamine and xylazine and placed in a

Figure 3-7. Soak testing result for 25 days: (a) Impedance changes in failed and successful
electrodes during the 25 days of soak testing. The impedance of the failed electrode increased
significantly at day 9. Optical images of (b) successful and (c) failed electrodes are shown. 3 out
of 30 electrodes were failed during the soak testing. The main cause of the failure was
delamination due to the insufficient protective sealing of Parylene on the periphery of metal
electrodes.
stereotaxic apparatus. Using sterile surgical procedures, a 3 – 4 cm incision was made in the skin
overlying the skull and a small region of bone was removed to expose V1. Using a glass pipette
and a pressure injection system, 1.0 μl (10 × 10^11 ~ 10 × 10^12 vector genome (vg)/mL) of the
virus solution (AAV-hSyn-hChR2(H134R)-mCherry; UNC Vector Core, NC, USA) was injected
at three to four equidistant locations. Injections occurred over a 2 – 3 minute period, and the
glass pipette was left in place for an additional 10 minutes to allow the vector to diffuse away
from the injection site. The cortex then was covered with Gelfoam and the skin overlying the

39

skull was sutured closed. The animals were allowed to recover from anesthesia, housed
separately, and given pain medication and antibiotics to minimize infection.

Figure 3-8. In vivo experiments: (a) mCherry fluorescence in a coronal slice of the transfected rat
V1. In vivo recording set-ups for (b) optical neural modulation and electrical recording of the
Opto-μECoG array and (c) the light-evoked extracellular recordings using the epidural LED
array.
For the device implant, the transfected (2 - 4 weeks after the injection) rat was re-anesthetized
and placed back into the stereotaxis. The developed Opto-μECoG array was surgically implanted
over V1 and the tests were conducted, as described below.
The capability of simultaneous optical neural modulation and electrical recording of the
Opto-μECoG array was investigated. In this case, the μ-LED array was placed on rat V1 (Fig. 37b) to optically modulate the neural activities with a light power density of ~ 1 mW/mm^2 and a
light pulse duration of 100 ms. ECoG signals induced by the LED light stimulation were
simultaneously recorded through the selected 15 ITO electrode channels. Recorded signals were
amplified using the RHA2000 amplifier array, then analyzed using MATLAB and Chronux
toolbox (Chronux.org). Fig. 3-9 shows representative results obtained when the μ-LED of
channel 6 was activated with an input voltage of 2.7 V at 1 Hz for 100 ms. The corresponding
light-evoked neural activity was translated into the time-varying energy distribution in the
gamma band (50 – 180 Hz) as shown in Fig. 3-8a. To study temporal and spatial characteristics
of the captured neural activity, a matrix of the average gamma band energies of the 15 channels
was generated every 50 ms after optical stimulation (Fig. 3-9c). In the first 50 ms window, the
40

highest gamma band power was observed at channel 6 where the light source was located. As
time passed, energy spread to neighboring channels. This preliminary result suggests that
epidural cortical stimulation by the LED array is capable of optically modulating cortical neural
activity.
To verify activation of transfected neurons located in deep cortical layers following
optical stimulation directed at the cortical surface, extracellular recordings were performed using
a carbon fiber probe (Carbostar-1, impedance range of 100 – 200 kΩ, Kation Scientific LLC,
MN, USA) at various cortical depths. To achieve this, a polyimide μ-LED array was prepared as
described in [30], only now with holes (~ 200 um in diameter) in spaces between μ-LEDs. Once
the μ-LED array was placed on V1, the carbon fiber probe was inserted into the hole next to the
activated μ-LED (Fig. 3-8b). While optically stimulating the superficial layer of V1 with an input
voltage range of 2.7 – 3.2 V and a light pulse width of 50 – 300 ms, the light-evoked neural
activity was monitored through the carbon fiber electrode controlled by a micropositioner
(Model 2670, David Kopf Instruments, CA, USA) at different depths. Recorded data were preprocessed using NeuroQuest, a MATLAB tool box for neurophysiological signal processing, and
light-evoked action potentials at different cortical layers of V1 were plotted (Fig. 3-9d) [127].
Clear spikes were observed at depths of 200 – 600 µm in response to the optical
stimulation, whereas no neural activity corresponding to the optical stimulation was observed at
a depth of 700 μm. After the in vivo experiment, the animals were perfused and the brain tissue
harvested for histological analysis to verify actual penetration depth and the thermal effect of
LED on brain tissue. This acute experiment revealed that epidural light stimulation using μLEDs was capable of evoking not only neurons in superficial layers (100 – 300 μm) but also
neurons in deep cortical layers (300 – 600 μm).

41

Figure 3-9. In vivo neural recording analysis results: (a) the spectrogram (50-180 Hz) of the
optically evoked ECoG signal recorded through the selected channel (Ch 6) where the activated
LED was located as shown in (b). (c) A series of a matrix of the average gamma band energy
with 50 ms interval. (d) Light-evoked extracellular recordings at different depths through the
carbon fiber electrode.

42

3.5 Conclusion
In this chapter, this author designed, fabricated, and characterized the Opto-μECoG array capable
of simultaneous optical stimulation and electrophysiological recording of cortical neurons. The
proposed hybrid neural interface had four unique features to address major challenges of existing
optogenetics-based BMIs: First, the multichannel LED array allowed the possibility of achieving
a truly untethered system crucial for chronic implants. Second, the optically transparent ITO
epidural electrodes enabled maximum exposure of the target cortical area for optical stimulation.
Third, integrated ITO electrodes and LED light sources carefully were arranged to avoid
redundant neural recordings and overlapping light stimulation due to closely-placed neighboring
electrodes and μ-LEDs, respectively. Finally, polymer packaging and the epidural placement of
the device on the cortex enabled long-term biocompatibility of the device and minimized risk of
potential tissue damage. Thermal characteristics of the device indicated the critical range of
stimulation parameters, such as input voltage and duration, so as to avoid brain tissue damage
from overheating. Optical and electrical properties of the device fulfilled design requirements for
an epidural optical neural modulator.
Short-term stability of the device was demonstrated using soak testing. In vivo acute
experiments verified the high-quality recording potential, and the possibility of chronic
implantation. This author also demonstrated for the first time that epidural LED light stimulation
could evoke neural activity not only within the superficial cortical layers, but also in deeper
cortical layers (to a depth of ~ 600 μm).
Although the thermal property of the array described in section 3.4.3 suggests that the
LED input voltage of 3.2 V for 300 ms results in local temperature increase of over 0.5 °C, no
significant tissue damages due to the thermal effect were observed. Since these thermal images

43

were collected without consideration of thermoregulation of cerebral circulation in the brain,
more systematic investigation must be done to determine actual thermal properties of the
implanted array.
Artifacts were observed during in vivo ECoG recording when the LED was powered on.
These artifacts are believed to be light-induced artifacts, also known as the Becquerel
photovoltaic effect. One possible solution to address this issue is to insert an extra layer of ITO
shield between electrode and LED arrays, which effectively can eliminate light-induced artifacts
and any electrical artifacts, as demonstrated by Zorzos et al. [128].
This novel opto-neural interface will permit multichannel, bi-directional interaction with
neurons at a level of sophistication not possible with existing optogenetics instrumentation, thus
lays the foundation for the development of advanced optogenetics-based BMIs. From system
design perspectives, this study has shown that LED input power of ~ 3.4 mW is sufficient for
activation of ChR2 expressing neurons. This makes it possible to drive the LED wirelessly with
suitable designs of inductive telemetry [76]. As an example of future chronic implants, it is
estimated that the 32-channel Opto-μECoG system would require a total input power of ~ 54.4
mW to permit optical stimulation through 16 LEDs simultaneously on one side of the bilateral
hemispherical cortex. Simultaneous recording of neural activities from 16 electrodes will
consume less than 9 mW, estimated from the power dissipation of Intan RHD 2000-series chips.
It is envisioned, approximately, that the proposed system will require a total power of < 65 mW
transferred through inductive coupling, with a coil separation of 1-3 cm in a rat model. Other
possible power sources may include small rechargeable batteries and energy-scavenging devices.
The effort to realize a fully implantable wireless neural interface is continued in Chapters 7 and
8.

44

CHAPTER 4
Three-dimensional polymer waveguides array for
optogenetics
4.1 Motivation
In the previous chapter, an epidural optical stimulating approach using the Opto-μECoG array, a
transparent μECoG electrodes array with integrated μ-LEDs, was introduced. While the 32 μLEDs array significantly improved the spatial resolution of the optical stimulation, its depth
resolution had a severe limitation. More, the depth control of the epidural optical stimulation
required relatively high input voltage and power to deliver light to deep cortical layers. Finally,
selective stimulation of the target layer was impossible with the surface light source.
This chapter presents the design, fabrication and characterization of two types of fully
implantable,

individually

addressable,

three-dimensional

(3-D)

micro-waveguides

(µ-

waveguides) array for optogenetics. Integration of individually addressable µ-LED chips with µwaveguides, shown in Fig. 4-1, enables minimization of the scattering of LED lights in the brain
tissue to achieve high spatial resolution, and precise light delivery to target neurons. The array
contains 32 embedded µ-LED light sources on a polyimide substrate. Polymer µ-waveguides are
fabricated separately on a PDMS substrate using a backside exposure technique and bond with
the multi-LED array using a shape-matching assembly. Physical dimension of the array is 2.5 ×
2.5 mm^2 with 16 (4 × 4 grid) channels per each hemisphere to meet specifications of bilateral
visual cortices in rats. Integrated light sources allow for the implementation of a truly untethered
system, crucial for chronic implant in freely behaving animals.

45

Figure 4-1. Conceptual illustration of the multi-LED array with 3-D polymer waveguides arrays
for deep brain optical stimulation.
The spatial resolution of the 3-D µ-waveguides array is further improved by proposing a
slanted waveguides array. The length-varying microneedle structure allows precise light delivery
to the target neurons in individual cortical layers simultaneously. Several approaches for making
a slanted microneedle structure have been reported, but they either required a specialized
machine and material [25], or lacked control of the microneedle size that determines light
coupling efficiency [129]. A new technique, “droplet backside exposure” (DBE) utilizing height
variance in the profile of a droplet, was developed to create the slanted microneedles array.
The DBE method uses surface energy differences between two adjacent contact areas of a
PDMS substrate to define the curvature of a droplet. As shown in Fig. 4-2, a dome-shaped SU-8
droplet with a designed base size can be formed on a patterned hydrophilic area (O2 plasma

46

treated-PDMS surface) with hydrophobic surroundings (intact PDMS surface). In a certain
volume range, the droplet is confined within the boundary between the hydrophilic and
hydrophobic regions, due to differences of surface energies, and this allows variant heights of
droplet/s by controlling the SU-8 volume. Then the slanted microneedles structure can be formed
in the dome structure using a backside exposure [130]. Details of the DBE method are reported
in Chapter 5.

Figure 4-2. Principle of the droplet backside exposure (DBE) method
This chapter first describes design specifications of 3-D waveguides arrays, followed by
devices fabrication. Next, the optical, electrical, and mechanical characteristics of the fabricated

47

device are investigated. Finally, the in vivo experiments are performed to investigate long-term
biocompatibility and reliability of the proposed devices.

4.2 Design
Several design objectives were considered for optical stimulation arrays. First, the μwaveguide must be able to deliver sufficient light to the desired cortical layer. The 3-D μwaveguides array was designed to target cortical layer IV in primary visual cortex (V1) of the
rat. Layer IV in V1 is an important area to study the connection between the visual cortex and
other deep brain structures, since V1 receives most visual input from the lateral geniculate
nucleus (LGN) through layer IV. Because layer IV in V1 is located at a depth range between ~
600 μm and ~ 1000 μm, the waveguide length was designed to be 800 μm. The slanted
waveguides array was designed with lengths ranging from 400 to 1000 μm to provide access to
multiple cortical layers (layers III to V) in the rat’s primary visual cortex (V1). For certain
applications of central nervous system (CNS) stimulation, delivering optical stimulation at
various cortical layers is critical. Moreover, to restore motor and sensory functions, peripheral
neuroprosthetic interface requires accessing multiple independent motor neuron subpopulations
in peripheral nerves or muscles.
Second, the μ-waveguide must be capable of evoking neural activity at the target cortical
region. This is related to coupling efficiency of the μ-waveguide. Typically 1mW/mm^2 of
throughput power is required to activate ChR2 expressing neurons. Butt coupling was utilized to
pair diffuse light sources (μ-LEDs) and μ-waveguides. A diffuse source means that the source
emits light in all directions and that the intensity of light is independent of direction. Butt
coupling is an attempt to capture emitted light before it diverges outside the fiber core radius, so
this simple coupling technique provides near optimal coupling efficiency with the size of the

48

fiber core smaller than the source radius [131]. To maximize coupling efficiency, the base
diameter of the waveguide was designed to be 300 μm, covering most of the surface area of the
light source (220 × 270 μm^2).
Third, the µ-waveguide should be mechanically durable to penetrate brain tissue without
breaking, while minimizing damage to brain tissue. A needle-shaped configuration - sharp tip
and large supporting base - was selected as a basic structure.
Finally, light sources must be addressable individually and distributed over the targeting
cortical area. Surface mounted μ-LED chips (220 × 270 ×50 μm^3, wavelength peak at 460 nm,
Cree® TR2227TM, Cree, NC, USA) coupled with the waveguide allow precise optical
stimulation in an individual or multiple spots simultaneously.
Within many available fabrication methods, polymer (SU-8)-based fabrication technique
utilizing backside exposure lithography was considered, because of the high refractive index of
SU-8, simplicity of fabrication technique, and biocompatibility suitable for bio-MEMS
applications [132]. While the adapted method allows length control with different size of mask,
independent control of dimensions (e.g. base and tip sizes, height) of the microneedle is not
possible [132]. Control of these parameters is critical in waveguide design, since these
parameters determine coupling efficiency, irradiance and total flux of the needle-shaped
waveguide [31]. The DBE technique allows to control lengths, tip and bottom diameters of
individual microneedles without sophisticated equipment and complex microfabrication.
Moreover, it is relatively simple with only slight changes from regular lithography processes,
and permits wafer-level microfabrication for large scale, high-density system integration.

49

Figure 4-3. Fabrication process of (a) multi-LEDs array, (b) 3-D μ-waveguides array, and
(c) slanted μ-waveguides array

50

4.3 Fabrication
To reduce fabrication complexity, the multi-LEDs array and the μ-waveguides array were
fabricated and calibrated separately. Then the system was assembled by polymer bonding of
individual components with SU-8. Detailed fabrication was divided into two steps and described
in the following sections (Fig. 4-3).
4.3.1 Multi-LEDs array fabrication
Fig. 4-3a depicts the core fabrication process flow of the multi-LEDs array assembly method.
Specifically, (a-1) A 3” silicon wafer was cleaned. (a-2) ~ 30 μm SU-8 layer was spun onto the
wafer and patterned as the mold for fabricating a PDMS stamp. (a-3) PDMS was poured over the
SU-8 mold to form the stamp, which contained cavities matching the shape of the LED chip. (a4) After curing for 40 mins at 95 °C, the PDMS stamp was peeled from the mold. (a-5) 32 LED
dies were aligned in the cavities of the stamp with metal pads facing outward. A substrate for
LED assembly was fabricated using Pyralux®AP (AP7163E, DuPont, DE, USA). (a-6) 3”
Pyralux® wafer was cut and cleaned, followed by Cu wet-etching. (a-7) Low melting point
(LMP) solder (melting point at ~ 62 °C, 144 ALLOY Field's Metal, Rotometals, Inc, CA, USA)
was applied on the contacts. (a-8) The PDMS stamp with the embedded LEDs was aligned to
match the LEDs’ metal contacts with pre-soldered receiver sites on the substrate. (a-9) The
substrate with the aligned PDMS stamp was heated on a hot plate at 90 °C for 30 secs. The
PDMS stamp was peeled off carefully after the substrate was cooled to 40 °C. The substrate,
with the attached LEDs, was then submerged into a hot acidic water bath (90 °C, pH of 2.0) for 1
min. This process permits fine adjustment of LED alignment and the formation of electrical
connection in a self-assembly manner [123]. After that, another layer of Parylene-C was

51

deposited to seal the entire structure, and contact vias were opened by oxygen plasma etching of
the Parylene.
4.3.2 A 3-D μ-waveguides array fabrication and assembly
Fig. 4-3b depicts the fabrication process flow of the 3-D μ-waveguides array fabrication and
assembly method. Specifically, (b-1) a 3” glass wafer was cleaned and put through a dehydration
bake. (b-2) a ~ 50 μm SU-8 layer was spun onto the wafer and patterned as the mock LEDs. (b3) well-mixed PDMS (Sylgard 184, Dow Corning, MI, USA) pre-polymer components in a 10:1
ratio were degased under vacuum until no bubbles appeared (20 ~ 30 mins). A thin layer of
PDMS was spin-coated on the SU-8 master to create cavities matching the LED shape. (b-4)
After curing for 40 mins at 95 °C, the PDMS substrate surface was treated with O2 plasma to
enhance adhesion between SU-8 and the PDMS surface. Then 800 μm SU-8 was spun on the
plasma-treated PDMS substrate and (b-5) patterned with the backside exposure to form the
microneedles. (b-6) After SU-8 development, the μ-waveguides array was polished by O2
plasma. (b-7) After releasing the PDMS membrane from the glass wafer, (b-8) the device was
assembled by polymeric bonding with shape-matching the PDMS cavities onto their
corresponding LEDs.
4.3.3 A slanted μ-waveguides array fabrication and assembly
Fig. 4-3c depicts the core fabrication process flow of the slanted μ-waveguides array fabrication
and assembly method. Specifically, (c-1) A 3” glass wafer was cleaned and dehydrated. A ~ 50
μm SU-8 layer was spun onto the wafer and patterned as the mock LEDs. (c-2) A thin layer of
PDMS (~80 μm) was spun on the SU-8 master to form cavities matching the shape of the LEDs.
(c-3) After the PDMS was cured at 95 °C for 40 mins, PR (S1813, Shipley, MA, USA) was
patterned on the PDMS substrate to expose 7-mm-diameter circles, followed by O2 plasma

52

treatment to convert the exposed lyophobic areas to the lyophilic ones. (c-4) After removing the
PR mask, ~45 μL SU-8 3005 (MicroChem Corp., MA, USA) was dispensed on top of the
plasma-treated PDMS surface using a micropipette. SU-8 droplets were pre-baked at 95°C for 12
hours in a vacuum oven, and (c-5) patterned with the backside exposure to form the
microneedles followed by post-baking at 95°C for 30 mins. (c-6) After SU-8 development, the
rough surface of the array was smoothed by O2 plasma treatment at power of 300W and gas
pressure of 0.5 Torr for 5 min and hard-baked at 150°C for 2 hours. (c-7) A thin layer of
aluminum was thermally evaporated as a reflective shield for the μ-waveguides array. The metal
layer on the tips then was removed using chemical wet etching, while the metal layer on the
sidewall of the waveguide was protected using an SU-8 mask. The SU-8 mask was formed by
dispensing SU-8 onto the region with the microneedles. During pre-baking, SU-8 reflowed due
to the capillary effect, resulting in a thick resist layer surrounding the microneedles and a thin
layer on the tip. After removing the thin SU-8 coating from the tip with oxygen plasma, the tip
was exposed by metal wet etching, followed by dissolving the SU-8 mask in the developer. (c-8)
The membrane carrying the microneedles array was released from the glass wafer, then (c-9) the
cavities in the membrane were aligned to the LED chips on the µ-LEDs array and bonded with
polymer adhesive.

4.4 Results
Fabricated prototype arrays and SEM images of the μ-waveguides are shown in Fig. 4-4. Several
experiments were conducted to characterize the 3-D μ-waveguide arrays in three different
aspects: optical, electrical, and mechanical properties.

53

Figure 4-4. Fabricated prototypes of (a) a μ-waveguides array and (b) the assembled array with
SU-8 waveguide. (c) SEM images of the SU-8 microneedle array fabricated on PDMS substrate.
(d) before and (e) after assembly with a slanted microneedle array with the flexible LEDs array.
(f) SEM images of slanted microneedle structures.
4.4.1 Optical properties
As important parameters for optical neural stimulation, the geometric effect of microneedles on
the irradiance and total flux of transmitted blue light were characterized, using a ray- tracing
method (TracePro®, Lambda Research Co., MA, USA). Normalized irradiance and total flux
were measured at a 100-μm distance from the microneedle tip, as the function of the tip size (Fig.

54

4-5a). With a small tip size, optical throughputs tend to spread out, resulting in divergent
irradiance, while a large tip size leads to concentrated irradiance with a confined output beam.

(a)

(b)
Figure 4-5. Simulation results: (a) irradiance measurements at a 100 μm distance from two tips
with diameters of 10 μm and 100 μm, and (b) normalized total flux measured at a 100 μm
distance from waveguide tips with different tip diameters (10 – 130 μm).

55

The coupling efficiency of the proposed slanted waveguide array was studied also. In this
case, several assumptions were made such as a typical LED radiation angle of ~ 60° and inserted
an 80-µm-thick PDMS layer (with the refractive index of 1.46 at 470 nm) between the LED and
the SU-8 waveguide (with the refractive index of 1.59 at 470 nm). Maximum coupling efficiency
of the waveguide with tip diameter larger than 100 μm was estimated to be around 80%, by
assuming the sidewall of the waveguide was a perfect mirror and all incidental light from the
LED entered the base of the waveguide. Simulation results suggest that the microneedle tip size
must be designed carefully to deliver sufficient optical throughput to the target area, while
permitting easy penetration of the tissue.

Figure 4-6. (a) Experimental setting to study optical throughput at the tip of waveguide, (b)
relative intensity of the blue light (450-495 nm) extracted from the captured image, (c) the zoomin view of the boxed area in (b). (d) Experimental setting to study optical throughput the μ-LED
chip without the waveguide, (e) relative intensity of the blue light (450-495 nm) extracted from
the captured image, (f) the zoom-in view of the boxed area in (e).

56

To characterize the light-scattering property of the array, the optical throughput at the tip
of the μ-waveguide in a scattering media (20 % gelatin) was studied. A 3-D μ-waveguides array
was placed on the gelatin media and images of the optical throughput at the tip of the single μwaveguide were captured. The activated LED was driven by 2.7 V resulting in a power
consumption of 3.4 mW, a temperature increase of 0.1 °C, and the light source irradiance of 1
mW/mm^2. Blue light spectra (wavelength range of 450 – 495 nm) were extracted from the
captured image, and the normalized relative light intensity (maximum intensity as 1) was shown
in Fig. 4-6b. A clear ellipsoidal scattering boundary (~ 600 μm in length and ~ 100 μm in width)
of the optical throughput, shown in Fig. 4-6b and c, indicated the waveguide effectively
delivered the optical stimulation at the targeting depth without stimulating intermediate layers, in
comparison with the large spherical scattering boundary of the μ-LED chip without the
waveguide in Fig. 4-6f.
4.4.2 Electrical properties
A minimal irradiance of 1 mW/mm^2 must be delivered to the target area to induce action
potentials. For surface stimulation, an input voltage of ~ 2.7 V was used to drive a single μ-LED
chip to achieve the required optical power output, resulting in a power consumption of 3.4 mW.
Considering typical coupling efficiency of the butt coupling (5 ~ 10 %) [131], the μ-LED chip
was driven by an input voltage of ~ 2.9 V to achieve an optical power intensity of 10 mW/mm^2,
resulting in a total power consumption of 17 mW and a temperature increase of ~0.4 °C.
4.4.3 Mechanical properties
Mechanical reliability of the μ-waveguides array was examined by a penetrating test. The μwaveguides array was inserted into and retracted from the 20% gelatin media 10 times, and
cracks or breakages on the μ-waveguide were examined microscopically. Among 10 samples, no

57

visible damage was observed, indicating mechanical rigidity of the μ-waveguide. The result
suggested possible array implantation with intact dura, which helps minimizing additional
complications caused by the dura-removal procedure.

(e)

Figure 4-7. Penetration profile of (a-d) the 3-D μ-waveguides array and (e) the slanted μwaveguides array sin different curvatures of gelatin surface.
Flexibility of this array was also examined by implanting it in gelatin (20% w/w) with
different surface curvatures. As shown in Fig. 4-7, the flexible PDMS substrate wrapped around
the curved surface conformally, while the waveguides penetrated perpendicular into the bulk
gelatin. This result demonstrated that a combination of the flexible substrate and the rigid
waveguide structure offers potential applications for not only cortical implants, but also spinal
cord and peripheral nerves implants. Specifically for peripheral nerve implants, each row of

58

waveguides can target different depths of axon bundles in the peripheral nervous system without
deforming nerves.

(a)

(b)
Figure 4-8. Evaluation of mechanical robustness of microneedle arrays using the axial loading
test: (a) from the top left, force measurement vs. displacement of the microneedle array with
different base diameters of 250, 200, 100, and 150 µm. (b) Needle failure force vs. base
diameter.
59

Also studied was the mechanical failure of microneedles with different base diameters,
based on the axial loading test described in [133], using arrays of four single-length
microneedles. These had various base diameters of 100, 150, 200, and 250 µm, constant length
of 900 µm, and tip diameter 25 µm. Previous study has shown that an interfacial needle area at
the tip plays an important role with respect to insertion of a microneedle into skin, while other
geometric parameters are less important [134]. According to this study, a 0.058 N/needle
insertion force would be required for a microneedle with tip diameter 25 µm, corresponding to
the interfacial area of 490 µm^2 [134].
During the test, each array was placed on a force gauge (Mark-10 BG5, Mark-10 Co.,
New York, USA) attached to a syringe pump (HA2000P, Harvard Apparatus, MA, USA). The
station pressed the microneedle array against a rigid glass surface at constant speed of 0.1 mm/s.
Displacement of the station measured from a laser distance sensor (OADM 20I6441/S14F,
Baumer Electric, Germany) and the corresponding force measured via the force gauge were
acquired using a dSPACE system (DS1104, dSPACE Inc., Germany).
Maximum force measured immediately before a reduction was seen as the critical force
required for needle failure due to buckling. Fig. 4-8a shows the failure forces for individual
microneedles with four different base diameters, estimated by dividing the total measured force
by the number of microneedles on each array. Ten sample arrays were measured for each base
diameter; the relationship between the failure force and the base diameter is shown in Fig. 4-8b
(the error bar indicating standard deviation). Results suggest that microneedles with base
diameters larger than 100 µm are mechanically robust enough to facilitate penetration into skin,
since average force of the microneedle with a 100 µm base diameter (0.12 N/needle) is above the
required insertion force (0.058 N/needle) in [134].

60

4.4.4 Long-term biocompatibility
Last, histological analysis was performed to evaluate long-term biocompatibility of the SU-8
microneedles, and to identify any evidence of trauma associated with the implanted device. All
animal surgical and postoperative care procedures were approved by the Institutional Animal
Care and Use Committee (IACUC) at Michigan State University.

Figure 4-9. Histological analysis for long-term biocompatibility evaluation: (a) 4 × 4 slanted
microneedle array was implanted on rat V1. (b) A coronal section (5 µm-thick and stained with
H&E) of the V1 after 6 months implant.
For device implantation, three rodent subjects (Sprague-Dawley rats: 250 – 400 g) were
anesthetized and placed in a stereotaxic apparatus. Using sterile surgical procedures, a 3–4 cm
incision was made in the skin overlying the skull and a small region of bone was removed to
expose primary visual cortex (V1). Dura mater was carefully removed, then a 4 × 4 slanted
waveguides array (300 – 1200 µm in length and ~ 100 µm in tip diameter) was implanted in V1,
as shown in Fig. 4-9a. Gel-form was applied on the craniotomy and the skin overlying the skull
was sutured closed. The animals were allowed to recover from anesthesia, housed separately, and
given pain medication and antibiotics to minimize infection. After recovery from surgery, the
rats were observed daily to detect any abnormal behavior due to the implant. None of the three
61

rats were used in the preliminary study showed noticeable behavioral changes. They then were
euthanized at 1-, 3-, and 6-month periods, and histological analysis was performed. After probes
were implanted, the brains were dehydrated in graded concentrations of ethanol, and embedded
in paraffin. The implanted hemisphere was sectioned into 5 µm coronal sections and stained with
hematoxylin and eosin (H&E). From Fig. 4-9b, mild glia cell formations and a small region of
tissue compression were observed around the microneedle tracks. This result indicated that the
flexible array combining a PDMS substrate with SU-8 microneedles would not cause significant
damage to the brain tissue, thus is suitable for long-term cortical implantation.

4.5 Conclusion
Design objectives, fabrication process, and the properties of 3-D μ-waveguide arrays for
optogenetics are described in this chapter. The 3-D μ-waveguide was designed to target layer IV
in V1 of the rat, and the fabrication was based on SU-8 μ-needle fabrication using the backside
exposure technique. Slanted μ-waveguides array was designed with lengths ranging 400 to 1000
μm to provide access to multiple cortical layers using a new fabrication method (DBE) for
slanted microneedles fabrication.
Comprehensive experiments studied the mechanical, optical, and biocompatible
properties of the as-fabricated arrays. The simulation result of optical throughput at the tip of the
waveguide was demonstrated and its scattering characteristic was analyzed. The presented
image-processing method was suitable for analyzing relative intensity distribution of the optical
throughput. Electrical power consumption to achieve irradiance of 1 mW/mm^2 at the
waveguide tip was estimated based on 10% coupling efficiency of the butt coupling. Mechanical
failure and in vivo animal tests validated that slanted microneedle arrays are mechanically robust
and biocompatible, suitable for chronic neural implants. Mechanical properties of the array were

62

demonstrated via the penetration test on gelatin media, and results suggested possible
applications of spinal cord and peripheral nerve implant.

63

CHAPTER 5
Varying-length polymer microneedle arrays
fabricated by droplet backside exposure
In the previous chapter, the slanted µ-waveguides array was introduced as an optical neural
interface for applications in optogenetics-based neural stimulation. A new fabrication method
was developed for making slanted microneedle structures monolithically on a flexible polymer
substrate, and this chapter presents a single-step backside ultraviolet (UV) lithography method,
namely droplet backside exposure (DBE). To demonstrate the feasibility of the DBE approach,
SU-8 microneedle arrays were fabricated on PDMS substrates. Microneedle length was
controlled by tuning the volume of the SU-8 droplet, utilizing the wetting barrier phenomenon at
a liquid-vapor-hydrophilic surface-hydrophobic surface interface. Experimental results supported
excellent repeatability and controllability of the DBE method for microneedle fabrication.
Analytical models and finite element method (FEM) were studied to predict microneedle
dimensions , which agreed with the experimental data. To verify versatility of the proposed DBE
method, a slanted hollow microneedles array was implemented.

5.1 Motivation
Microneedle based on microelectromechanical systems (MEMS) technologies have been an
attractive topic for decades, since they can serve as an interface between an organic system and
an external micro device [135]. Such interfaces are capable of either delivering a liquid
substance such as drug, protein, and nano-particles into the organism, or extracting biofluids
from the organism for analysis [136]. In addition, microneedles can be used as microelectrodes

64

to interface with the nervous system for neural recording and stimulation [137]. Recently,
optically-transparent microneedles made of glass [25] and polymer [30] were reported as µwaveguide for optical neuromodulation based on optogenetics, which extend the potential of
microneedles.

Figure 5-1. Principle of the proposed droplet backside exposure (DBE) method.
Depending on their geometries, typically, microneedles can be classified into two groups:
in-plane microneedles and out-of-plane microneedles. In most cases, in-plane microneedles are
fabricated with conventional silicon MEMS techniques, and can be used for precise drug
delivery to a target region or tissue of the body [135]. Contrarily, out-of-plane microneedles
require complex fabrication processes, and can be made over a large area and with a high yield
and excellent repeatability. Out-of-plane microneedles have been used widely in hypodermic
drug delivery applications [138] and cortical neural interface [1]. Accordingly, this author
focuses discussion on microfabrication techniques for making out-of-plane microneedles.
To date, a wide variety of materials, including silicon, metal, glass, polymers, and
biodegradable materials, have been used to fabricate out-of-plane microneedles, depending on a
target organism and/or a specific application of the interface [136]. Various microfabrication
methods also have been developed based on different building materials. For example, silicon-

65

based out-of-plane microneedles typically were fabricated using either reactive-ion etching (RIE)
[136] or a combination of anisotropic wet etching, deep-RIE, and anisotropic bulk
micromachining processes [139], [140]. Metal microneedles were developed in [141], using
electroplating onto a seed layer sputtered on polymer or silicon micromolds. Moreover, glass
microneedles were made using a conventional micropipette puller, or micromachining then
etching a thick glass substrate [25]. However, these fabrication approaches are labor intensive
and costly. With superior biocompatibility and low-cost manufacturing capability, polymeric
materials have become ideal material candidates for constructing microneedles [129], [133],
[138]. Previously- reported methods include soft lithographic replication by melting polymer into
a master mold, as well as backside exposure using photoresist such as PMMA and SU-8 [129],
[138].
Despite advances in microfabrication, microneedle arrays fabricated with the
aforementioned methods typically have a uniform length, primarily designed for applications of
hypodermic drug delivery. These arrays are unsuitable for specific applications requiring
comprehensive access to different locations of an organism, where naturally have distinctive
functionalities. As an example, to restore motor and sensory functions, peripheral
neuroprostheses interfacing with peripheral nerves or muscles require selective stimulation of a
large number of individual muscle groups with graded recruitment of force. In addition to
peripheral neural interfaces, BMIs utilize the cerebral cortex as a major target because of its
accessibility and its key role in cognitive, sensory perception, and motor control [1]. A
distinguishing feature of the cerebral cortex is a layered structure, where different cortical layers
contain a characteristic distribution of neuronal subtypes and connections with other cortical and
subcortical regions. For these applications in neuroprosthetic interfaces, a varying-length or

66

slanted microneedle array is necessary. One such example is the Utah Slanted Electrode Array
(USEA), which has been demonstrated to provide comprehensive access to multiple independent
motor neuron subpopulations [142]. While there is a critical need for slanted microneedle arrays,
fabrication techniques for making such arrays have not been studied extensively. Currently, only
some methods have been reported. In particular, the USEA was fabricated by dicing and etching
bulk materials, such as silicon or glass [25], [137]. However, this method relies on specialized
micromachining tools and bulk materials, which may limit its adaptation by other researchers in
this field. Stacking in-plane microneedles with different lengths is another way to build a
varying-length microneedle array, but this approach requires a complex assembly process [133].
While polymer-based backside exposure method allows for precise length control with different
mask aperture sizes [143], independent control of microneedle dimensions (length, tip and
bottom diameters) is not possible.
To address these challenges, a unique fabrication technique, called droplet backside
exposure (DBE), has been proposed for making polymer-based (SU-8) varying-length
microneedle structures. Microneedles with different lengths can be achieved using the height
variance in the profile of a droplet structure through rapid, single-step UV lithography (Fig. 5-1).
This DBE technique enables the independent control of the length, tip and bottom diameters of
individual microneedles, without using a specialized machine or complex fabrication techniques.
The following sections will detail the principle of the proposed method, the DBE fabrication
steps, and its adapted processes for making slanted hollow microneedle array. Experimental
results have demonstrated the repeatability and versatility of the proposed fabrication method.

67

Figure 5-2. An analytical model for designing slanted microneedle arrays, approximated based
on the geometry of an ideal partial sphere droplet: (a) a diagram of an ideal partial sphere, or a
spherical cap; (b) the ideal droplets, formed on the hydrophilic surface, with a minimum CA
(αmin) and a maximum CA (αmax), respectively; (c) the relationship between two partial spheres,
V1 and V2 .

68

5.2 Methods
5.2.1 Principle of the DBE
The proposed DBE technique takes advantage of a wetting barrier phenomenon occurring at a
four-phase interface [144]. Generally, the shape of a liquid droplet on a homogenous solid
substrate is primarily governed by the surface free energy of the substrate. However, for a fourphase interface such as a vapor-liquid-hydrophilic surface-hydrophobic surface interface, the
spreading of the liquid is restricted by the wetting barrier formed at the boundary between the
hydrophilic and hydrophobic regions. Consequently, the contact angle (CA) of the droplet can
vary within a certain range, with a maximum value corresponding to the equilibrium CA of the
liquid on the hydrophobic surface, as shown in Fig. 5-2b.
If it is assumed that the droplet is a perfect partial sphere with a pre-defined diameter, a
relationship between the CA and the volume of the droplet can be estimated. Based on this
principle, an analytical model was derived to determine the maximum volume of the liquid
sustained in a pre-defined pattern and to predict the length variation of microneedles. At
equilibrium, the volume of the liquid can be calculated using the following equation:

(

(

)

(

))

(1)

where V is the actual volume of the liquid delivered to the substrate and α is the equilibrium CA
of the droplet. The diameter of the interface between the droplet and the hydrophilic surface,
denoted as d in Fig. 5-2b, is pre-defined so the relationship between R and d can be derived as:
(

)

69

(2)

By substituting R with d in Eqn.1, the liquid volume can be rewritten as a function of the
pre-defined interfacial diameter and the equilibrium CA (α) of droplet.
(

)(

)

(3)

Once the CA of the droplet is determined by the volume, the profile of the droplet can be
estimated. hmax is defined as the maximum height of the droplet, which can be estimated using a
trigonometric relationship of R and α shown in Fig. 5-2a.

(

)

(4)

Then the length (h) of a microneedle formed at a certain distance (b) from the center of
the droplet can be calculated using a relationship between two partial spheres V1 and V2 in Fig.
5-2c, as given below:
√( )

( ( )

) ( )

(5)

In this study, the choice of liquid was SU-8 for polymer-based microneedle fabrication.
Particularly, SU-8 3005 (with 50% resin solid, MicroChem Corp., MA, USA) was used, because
its low viscosity (65 cSt) allows accurate volume dispense. Polydimethylsiloxane (PDMS) was
selected as the flexible substrate material, because the innate PDMS surface is hydrophobic to
the primary solvent of SU-8, cyclopentanone, and effectively can be modified to super
hydrophilic through an oxygen (O2) plasma treatment. Among a same array, the dimensions of a
microneedle, including its tip diameter, base diameter, and length, can be designed precisely
based on the above analytical model. More specifically, base diameter is determined by the

70

diameter of mask aperture, while tip size is associated closely with the inclined angle of the
tapered microneedle and the thickness of the SU-8 coating [143]. Since the inclined angle is
mainly controlled by the separation between the absorber on the photomask and the top surface
of the SU-8 droplet [129], the only adjustable parameter is the thickness of the SU-8 layer.
Because this thickness varies in a predictable manner across the SU-8 droplet, the length and tip
diameter can be controlled by tuning the distance from the mask aperture to the center of the
droplet.

Max. Height
(mm)

1.2
Without Gravity
With Gravity

0.9
0.6
0.3

d=3mm

0.0

0

Max. Height
(mm)

2.0

2

4

6

Without Gravity
With Gravity

1.6
1.2
0.8
0.4

d=7mm

0.0
0

10

20

30

Volume (L)

40

50

60

Figure 5-3. Finite element stimulation result of gravity effect on the maximum height of SU-8
droplets with different base diameters.

71

5.2.2 Volume shrinkage and gravity effect
Factors that must be considered to achieve more accurate length estimation: the gravity effect of
the droplet and the volume shrinkage caused by solvent evaporation and cross-linkage. When a
droplet is placed on a horizontal surface, the droplet shape will be deformed by the gravity force.
This effect depends on the size of the droplet as well as the given materials. Since the proposed
method assumed that the droplet is an ideal partial sphere, the analytical estimation would not be
accurate for a large-size droplet with significant gravity effect. Vafaei at el. reported an
analytical model to estimate the droplet geometry under gravity effect in [145], but this model
requires experimental measurements of the maximum height, CA radius, and equilibrium CA of
the droplet.
To study the gravity effect, finite element simulation in COMSOL Multiphysics 4.3
(COMSOL, Inc., MA, USA) was performed using the Laminar Two-Phase Flow (Level Set)
model. Two different diameters of the plasma-treated hydrophilic circle (d), 3 mm and 7 mm,
were studied. The equilibrium CAs of SU-8 was about ~ 10° on the plasma treated PDMS and ~
58.7° on the untreated PDMS, measured by a CA analyzer. The simulation used 7.95 mm^2/s for
the kinematic viscosity of SU-8 3005 at an elevated temperature (between 60 °C and 95 °C)
[146]. Two simulation parameters were assumed: the SU-8 density of 1.075 g/cm^3 and the
surface tension of ~ 48 mN/m [146]-[148]. The droplet was defined as a hemisphere at t = 0 sec,
with a radius varied from 0 mm to 1.4 mm in the case of a 3 mm base diameter and from 0 mm
to 3.4 mm in the case of a 7 mm base diameter, in 10 µm increments. During the simulation, the
droplet continuously spread over the heterogeneous substrate until its CA reached the maximum
equilibrium CA (~ 56° for SU-8 on untreated PDMS). Then the maximum height of the droplet
for each individual radius was recorded before and after applying the gravity effect, and plotted

72

as a function of the droplet volume (Fig. 5-3). Simulation results suggest that the gravity effect
on the maximum height of the droplet was negligible when droplets had a small base diameter
and low volume sizes. For a large base diameter of 7 mm, when the droplet volume was greater
than 25 µL, the gravity effect caused a significant reduction in the maximum height by over 10
%. Compared to the experimental data, the current analytical model is considered valid with an
error of smaller than ~ 8.0%, for a droplet with a base diameter of ≤ 7 mm and a volume of ≤ 25
µL. Slight oscillation of the SU-8 surface was observed during the transient analysis of the Level
Set model, which was assumed to have a negligible effect on the droplet shape.
Volume shrinkage due to solvent evaporation and SU-8 cross-linking was modeled by
assuming that the droplet still remains a spherical shape after pre- and post- baking. The volume
shrinkage factor of ~ 7.5 % was estimated based on empirical analyses where the contact angles
and the volumes of one hundred droplets were measured and compared before and after baking.
The empirical result is consistent with the findings reported by Lorenz et al. [147]. Applying this
factor in Eqn. 1 results in a new formula for calculating the actual droplet volume, Vact, as
expressed in Eqn. 6.
(

)(

)

(6)

where S is a percentage of solid in SU-8 solution multiplied by 0.925.

5.3 Fabrication methods
5.3.1 Basic DBE fabrication process
Fig. 5-4a depicts the core fabrication process flow of the DBE method. Specifically, (a-1) PDMS
pre-polymer components (Sylgard 184, Dow Corning, Midland, MI) were well mixed in a 10:1
ratio, and the mixture was degased under vacuum until no bubbles appeared (20 ~ 30 min).

73

PDMS was then spun on a clean 3” glass wafer. After the PDMS was cured for 40 min at 95 °C,
photoresist (PR, S1813, Shipley, Marlborough, MA) was spun on the PDMS.

Figure 5-4. The core fabrication process of the DBE method and its variation: (a) the core DBE
fabrication method; (b) a slanted hollow microneedle fabrication process flow.

74

Circular patterns with various diameters (3 ~ 7 mm) were formed in the PR layer through
photolithography, followed by an O2 plasma treatment to define hydrophilic areas for containing
SU-8 droplets. (a-2) After removing the PR masking layer, SU-8 3005 was dispensed on top of
the plasma treated PDMS areas using a micropipette. Then SU-8 droplets were pre-baked at 95
°C for 12 hours in a vacuum oven, and then (a-3) patterned with the backside exposure to form
slanted microneedle arrays. (a-4) After SU-8 development, the slanted microneedle arrays were
polished with O2 plasma etching. (a-5) Finally, the flexible microneedle arrays were released
from the glass wafer.
5.3.2 A slanted hollow microneedle array fabrication process
The basic DBE method is fully adaptable to the fabrication of hollow microneedles, by
constructing a multi-layer masking structure for the backside exposure process. In this case, a
metal mask was first patterned on the top side of a glass wafer, then a second mask was placed
on the bottom side of the same wafer after pre-baking the SU-8 droplet. Due to the different
separations between the two masks and the bottom of the SU-8 droplet, the outside wall of the
microneedle will have a larger inclining angle, while the inside wall will have a smaller angle,
resulting in a hollow needle structure [129].
Fig. 5-4b depicts the fabrication process flow for making slanted hollow microneedles
using the modified DBE method. Overall fabrication flow was similar to the core DBE process
introduced in Fig. 5-4a. Additionally, two important steps were incorporated, to form channels
through the PDMS substrate and the microneedles. Briefly, (b-1) a 3” glass wafer was cleaned by
IPA, acetone, and deionized (DI) water, then a metal (Al) mask was patterned on the wafer. (b-2)
Approximately 80-µm-thick PDMS was spun (800 rpm for 40 sec) and cured for 40 min at 95
°C. Then another layer of metal was deposited and patterned as a RIE etching mask. (b-3) A mi-

75

Figure 5-5. (a) Dispensed SU-8 droplets on PDMS. (b) Example of CA measurement. SEM
images of (c) the inclining profile of the microneedles with a 35.2° CA, and (d) the aerial view of
a circular microneedle array fabricated using a 20 μL droplet. (e) and (f) SEM images of a
hollow microneedle array.

76

xture of CF4 and O2 was utilized as the reactive gas to etch through the channels in the PDMS
layer. (b-4) After removing the RIE etching mask, the subsequent processing steps were the same
as the steps (a-2) to (a-5) in the core fabrication process.

5.4 Results
Prior to microneedle fabrication, the equilibrium CAs of the SU-8 on both intact and plasmatreated PDMS substrates were measured using a CA analyzer, as shown in Fig. 5-5b, to guide the
design of microneedle arrays. The intrinsic CAs on hydrophobic,

, and hydrophilic,

,

PDMS substrates were ~ 58.7° and ~ 10°, respectively. With an interfacial diameter (d) of 7 mm,
the maximum height of the droplets (

) was estimated to range from 306 μm to 1968 μm,

corresponding to the droplet volumes of 18 μL to 38.7 μL.
Fig. 5-5c shows a slanted microneedle array, which was fabricated with the droplet
diameter of 7 mm using the proposed DBE method. This array had an overall slope of ~ 35.2°,
resulted from the droplet volume of 20.4 μL. Circularly arranged microneedles, fabricated from a
20 μL droplet, are shown in Fig. 5-5d. The SEM images of the as-fabricated hollow microneedle
array were given in Fig. 5-5e and f.
Consistency of the DBE method was first investigated empirically with a fixed droplet
volume. In this study, ten 25 µL SU-8 droplets were dispensed on O2 plasma treated PDMS
circles 7 mm in diameter, resulting in an average CA of 41.5°. Ten waveguide arrays were
fabricated from these droplets using the basic DBE method as described previously. For the
backside exposure, it has been found that the tip size of the tapered microneedles increases as the
separation between the mask and the top SU-8 layer decreases [129].
To achieve a uniform tip size of microneedles, the aperture diameter of the mask was
changed from 140, 120, 100, to 80 μm, corresponding to the distance from the droplet center of

77

780, 1440, 2130, and 2830 μm. Average heights of fabricated microneedles were measured and
plotted in Fig. 5-6a, depending on the estimated droplet profile. Waveguide lengths were close to
estimated values with standard deviations of less than 5%, demonstrating the excellent
repeatability and reliability of the proposed DBE method.
Further, the relationship between droplet volume and microneedle length was studied.
With the same base diameter of 7 mm, the droplet volume varied from 10 to 27 μL with a 0.5 μL
increment, and the same mask apertures described above were used. Ten arrays of microneedles
were made for each volume size, and their lengths were measured from the SEM images and
compared with the theoretical values estimated from Eqn. 6. The measured heights of the
microneedles fitted well with their estimated lengths with standard deviations of less than 5 %, as
shown in Fig. 5-6b.
To further investigate the controllability of microneedle shape in the DBE method, the tip
and bottom diameters of the samples fabricated in the previous section were measured and
plotted in Fig. 5-6c. The preliminary results showed that the base diameter of the microneedle
was mainly determined by the diameter of the mask aperture and did not change with different
droplet volumes, while the tip size showed a strong correlation with the change of droplet
volume.

5.5 Conclusion
The DBE method for making slanted solid and hollow microneedles was designed, developed,
and characterized in this chapter. The results have shown that the three key parameters of the
microneedle (the length, tip and base diameters) can be controlled precisely by manipulating the
droplet volume and the mask design (the aperture diameter and the distance from the center). The
measurements of these parameters were consistent with the analytical estimation, demonstrating

78

(a)

(b)
Figure 5-6. Dimension control of the DBE method: (a) comparison between the estimated
heights and the measured heights of the fabricated microneedles with the standard deviations of
less than 5%. (b) Estimated and measured heights of microneedles formed at different distances
from the center of the droplet: 780 μm, 1440 μm, 2130 μm, and 2830 µm. (c) Dependency of
base (left) and tip (right) size on SU-8 droplet volume. This experiment tested microneedles
fabricated using four aperture diameters of 140, 120, 100, and 80 μm, at 780, 1140, 2130, and
2830 μm distance from the center of the droplet, respectively, with 24 different volumes ranging
from 10 µL to 27 µL.

79

Figure 5-6 (cont’d)

(c)

the repeatability and controllability of the DBE method for fabricating varying-length
microneedle.
In addition, flexibility and practicability of the DBE technique were proven with two
adapted processes, by which a slanted hollow microneedle array and a slanted 3-D µ-waveguide
array coupled with µLEDs, presented in Chapter 4, were realized.

80

CHAPTER 6
Development of a cortically based hybrid visual
prosthesis
6.1 Cortically based visual prosthesis
Based on works presented above, this author aims to develop a cortically based opto-epidural
visual prosthesis for therapeutic management of all forms of blindness. Blindness has affected
more than 40 million people worldwide [149] and approximately 1.2 million Americans [150]. In
the United States, the annual cost of healthcare for individuals with visual impairment and
blindness is over $5.48 billion and the total economic impact is estimated at nearly $51.4 billion
annually [151]. Unfortunately, a solution to cure severe vision loss and blindness has not been
found. Visual prostheses provide a potential clinical intervention to restore basic visual function
of blind individuals through electrical stimulation of residual neurons in the visual pathway.
Among many visual prostheses, the retinal prosthesis represents the most advanced technique
and several such devices have entered clinical trials [152]-[160]. While significant improvement
in visual function has been proven with the existing retinal prosthetic devices, these are only
effective in treating blindness with functioning retina (e.g. age-related macular degeneration and
retinitis pigmentosa). Alternative approaches are needed for the treatment of a large blind
population where the retina and/or the optic nerve are severely damaged as in glaucoma (second
leading cause of blindness (12.3 %) in the world), diabetic retinopathy (4.8 %), and traumainduced blindness [161].

81

A cortical prosthesis bypasses the retina and the optic nerve, thus offers a general
therapeutic opportunity for all causes of blindness. Early attempts at the cortical visual prosthesis
were made by placing epidural electrodes on the visual cortex, electrical stimulation of which
elicited phosphenes [20], [162]. Major shortcoming of the epidural stimulation is the high current
needed to elicit phosphenes (in mA), resulting in poor spatial resolution, discomfort, and even
focal epileptic activity [163], [164]. Intracortical electrodes can enable a reduced current
threshold and improved spatial resolution of phosphene generation [165]-[167], but the invasive
implantation of the penetrating probes increases the risk of stroke, hemorrhage, and infection. As
mentioned in Chapter 2, inevitable limitations of electrical stimulation including non-specificity,
unpredictable current pathway, electrical artifact, and electrode degradation, also increase the
demand for a new technology [21].
This author hypothesizes that epidural optical stimulation of a large population of ChR2
expressing neurons in the primary visual cortex (V1) can induce phosphenes that lead to artificial
prosthetic vision. This hypothesis is inspired by the electrically-based cortical visual prosthesis
and formulated based on recent studies showing that optical stimulation of ChR2 expressing
neurons in monkey V1 induced phosphenes and visuomotor behavior [168], [169]. Despite this
behavioral evidence, little is known about spatial and temporal characteristics of the V1 neural
activities in relation to the light-evoked phosphenes that are the key steps toward realization of a
cortically-based opto-epidural visual prosthesis for treating blindness.
To address this knowledge gap, the hybrid neural interface was utilized for multichannel,
bidirectional interaction with a large population of neurons in the V1 to explain the underlying
correlation between light-evoked neural activities in the ChR2 transfected V1 and phosphene
induction using advanced signal processing techniques. In the preliminary work discussed in

82

Chapter 3, this author successfully had shown that the Opto-µECoG array enabled simultaneous
optical stimulation and recording of neural activities with significantly improved spatial
resolution. Using the Opto-µECoG array, in vivo animal experiments based on rodent models
have been designed to study the correlation between neural activities in the V1 and light-evoked
phosphenes. Such experiments include 1) stimulus-driven ECoG data analysis; 2) a chronic
device development for freely behaving animal experiment; 3) a behavioral test to analyze ECoG
data in response to external visual stimuli; and 4) a behavioral test to examine the ability of
optical stimulation of ChR2 expressing neurons to induce phosphenes. Several specific tasks
must be done to prove the principal hypothesis, and details of the each task are discussed in the
text below.

6.2 Major tasks for development of the hybrid visual prosthesis
6.2.1 Stimulus-driven ECoG data analysis
This author anticipated that multichannel ECoG data contain spatial and temporal information of
neural ensemble encoding in response to a specific stimulus. These time-dependent spatial
activation patterns are named ECoG snapshots. Based on this idea, an acute animal experiment
was conducted to collect visual stimuli-driven ECoG data over the entire V1 through a highdensity ECoG microelectrode array. Analysis of these data will furnish better understanding of
underlying correlation between visual stimulus and the ECoG snapshots. Preliminary analysis
results of stimuli-driven ECoG data are presented later in this chapter.
6.2.2 Chronic versions of the hybrid neural interfaces
Since the perception of artificial visual sensation in the animal model only can be verified in
freely behaving animal experiment, development of chronic wireless neural interface is a critical
step toward project success. Two different types of chronic devices have been developed based

83

on the flexible µLED array. First, this author is working on the extension of the slanted
waveguide array to make wireless hybrid neural interface using the inductive powering system
based on switch-capacitive stimulation (SCS) system developed by the collaborators at Georgia
Institute of Technology [170] . This author and the collaborators have designed the SCS system
fully integrated with the LED array on a single flexible PCB board, and short-term and long-term
reliability tests of the SCS-based wireless system is ongoing at time of this writing.
Another chronic device has been designed for a freely behaving experiment for mice.
This battery-powered µ-LED array was designed for a small animal model that cannot carry an
external device. Three versions of the device are available depending on requirements of
experiments, and in vivo chronic animal experiment is in progress to evaluate long-term
reliability test of the devices. Current developments of the chronic wireless hybrid neural
interfaces are presented in Chapters 7 and 8.
6.2.3 Design of a behavioral animal experiment
As a proof of concept, a simple behavioral test was designed to verify perception of phosphenes
using a rat model as shown in Fig. 6-1. First, in the visual cue association test, a single visual cue
will be given to the rat. Within a certain time, the rat has to press a lever to receive a reward. If
the rat does not press the lever after the cue, then no reward will be given. Once the animal is
able to associate with the vision cue and the lever press task, then the Opto-µECoG array is
implanted on the V1. After recovery from implant surgery, the subject will be back to the
training protocol. This time the vision cue will be replaced with epidural light stimulation using
the Opto-µECoG array, and the behavioral score will be recorded. If the rat presses the lever, it is
possible that the epidural light stimulation induces phosphenes. Once the first test is successful,
via more complicated pattern discrimination test will be tested.

84

Figure 6-1 Proposed behavioral animal experiments to test artificial visual perception using
cortically optical neural stimulation of the rat V1.
In this test, two distinguished vision cues are given, and each cue is associated with a
lever. For example, cue 1 (horizontal line) is associated with lever 1, and the rat must press lever
1 to receive a reward when cue 1 is present. Similarly, cue 2 (vertical line) is connected with
lever 2, and pressing only lever 2 will deliver a reward. During the training, neural activities
(ECoG snapshots) during presence of cue 1 and cue 2 will be recorded, and data will be analyzed
to find the stimulation parameter to reproduce the neural activity induced by the two different
vision cues. After estimating the stimulation parameter during training session, vision cue will
be replaced with epidural light stimulation using the Opto-µECoG array, and the behavioral
score will be recorded. This author is aware of the high-risk nature of the new approach and
proof-of-concept animal model studies, because it is not known whether simultaneous optical
modulation of multiple neural populations can evoke visual sensation. So the chronic device will

85

be first tested in a well-established reflexive behavioral model, termed the gap-prepulse
inhibition of the acoustic startle (GPIAS). More details about the behavioral experiments are
discussed in Chapter 8.

6.3 Investigation of phase-locked neuronal oscillation
As a part of the major tasks for development of the hybrid visual prosthesis, stimuli-driven
neural activity was first studied. The effects of optical stimulation in optogenetics have been
studied in many aspects such as synaptic connections, behavioral responses, and corticocortical
and corticothalamic connections [171]-[173]. However, in-depth understanding of characteristics
of an optically-evoked large population of neuronal activity is not available. Neuronal
oscillation, a fundamental component of brain function, is a synchronized synaptic activity
across the local population of neurons. It can be interpreted into specific spatiotemporal patterns
revealing how information flows through the brain [174]. It is believed that neuronal oscillation
plays a critical role in large-scale functional integration enabling communication in large
neuronal networks [175]. In general, a synchronized activity of large neuron populations across
broad brain regions generates slower oscillations, while smaller localized neuronal assemblies
are associated with faster oscillations [176]. ECoG recordings measure those brain activities
directly with relatively higher spatial resolution, broader bandwidth, and higher amplitude,
compared

to

noninvasive

methods

such

as

electroencephalography

(EEG)

and

magnetoencephalography (MEG) [174].
Using the Opto-μECoG array, neural oscillations induced by natural and optical neural
stimulations in an optogenetic rat model were studied. Phase synchronization of the recorded
ECoG signals were analyzed using a new phase estimation method, based on the reduced
interference distribution time-frequency phase synchrony (RID-TFPS) measure [177] and a

86

conventional instantaneous phase estimation based on the Hilbert Transform. The new method
offers a phase estimation with uniformly high time-frequency resolution that can be used for
defining time and frequency dependent phase synchrony. In the following sections, a short
description of the RID-Rihaczek distribution-based phase synchrony measure is introduced.
Next, in vivo experiments for stimuli-driven ECoG recordings are presented. Finally, data
analysis results are presented and discussed.

6.4 Phase synchrony estimations
Existing measures to quantify neural connectivity from multichannel physiological data can be
categorized broadly into linear and nonlinear measures [178]. Linear measures such as temporal
correlation, spectral coherence, directed transfer function (DTF), partial directed coherence
(PDC), and Granger causality, suffer from the assumption that underlying signals are stationary
and the nature of the interactions between neuronal oscillations is linear. Nonlinear measures, on
the other hand, attempt to address this limitation through phase synchrony [179], mutual
information, and generalized synchronization measures. Recently, Aviyente et al., introduced a
new time-frequency phase synchrony measure based on a complex energy distribution, referred
to as Reduced Interference Rihaczek Distribution (RID-Rihaczek) as follows:

(
where

(

)
(

)

∬

(

(

)

)

(

) (

)

(

)

(7)

) is the kernel function (in this case the Choi-Williams kernel), which is

used to filter out the cross-terms due to multiple signal components in the original Rihaczek
distribution, and

∫ (

)

(

(
)

) is the kernel function of the Rihaczek distribution [177]. (

)

is the ambiguity function of the signal. The spread of the

87

window

is inversely proportional to

and determines the frequency of time-varying phase

estimates.
Based on this new time-frequency distribution, the phase difference between two signals
can be defined as

(

)

[|

(
(

)
)||

(
(

)
)|

]

(8)

Once the phase difference is quantified, phase synchrony can be computed. In this work,
this author focuses on the intertribal phase synchrony that quantifies the phase locking to the
response across trials. A phase locking value (PLV) is defined as

(

)

|∑

(

(

))|

(9)

where N is the number of trials and is the time-varying phase estimate between a reference trial
and the kth trial/realization.
This novel measure has been shown to provide better time-frequency resolution in
tracking the time-varying phase synchrony and increased robustness in noise compared to
existing methods [177].

6.5 In vivo recordings
In vivo acute animal experiments were conducted to collect stimuli-driven ECoG recordings. All
procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at
Michigan State University. Prior to the acute experiments, rodent subjects (Sprague-Dawley rats:
250-400 g) had neurons in their V1 transfected with ChR2 to enable light sensitivity. For device
implantation, the transfected rat (2 - 4 weeks after virus injection) was anesthetized with a
mixture of ketamine/xylazine and placed in a stereotaxic apparatus. The Opto-μECoG array was
surgically implanted over V1 and specific tests were conducted as described below.

88

Figure 6-2. In vivo tests with (a) a natural visual stimulation and (b) a cortically optical
stimulation.
Visual stimulus-driven ECoG signals were recorded first with a custom-made visual
stimulator. Two green LEDs (a peak wavelength at 560 nm) were embedded inside a plastic
mask to shield ambient light from rat eyes. The mask was installed on a stereotaxic (Fig. 6-2a)
and a visual stimulus was delivered to the rat’s left eye with 10 ms duration in a 1 s trial. ECoG
signals were recorded from both primary visual cortices (V1) through the selected 23 channels
(11 channels from right V1 and 12 from left V1) of the Opto-μECoG array.
The second experiment was performed to record cortically optical stimulus-driven ECoG
recordings. In this case, the blue μ-LED (a peak wavelength at 460 nm) array was placed on the
rat V1 (Fig. 6-2b) epidurally to optically modulate the neural activity with a light power density
of ~ 10 mW/mm^2 and a light pulse duration of 10 ms. ECoG signals induced by the μ-LED
light stimulation were recorded simultaneously through 15 selected ITO electrodes. The recorded
signals were amplified using an RHA2000 amplifier array then analyzed using MATLAB.

89

Figure 6-3. Instantaneous phase measure across trials based on Hilbert transform of the ECoG
signals bandpass filtered at 4 – 8 Hz: (a) Visual I (106 trials at channel 4), (b) Visual II (114
trials at channel 4), and (c) Opto I (38 trials at channel 10). The black line on (a), (b), and (c)
indicates stimulation ON. (d) Instantaneous phase and their corresponding color codings are
shown.

6.6 Results
Three types of ECoG signals were recorded with the natural visual stimulation and the cortically
optical stimulation via the Opto-μECoG array. The first dataset (Visual I) was recorded from the
right V1 with the left eye stimulation (green LED) with 10 ms duration in a 1 sec trial (total of
106 trials). The second dataset (Visual II) was recorded from the left V1 with the same
stimulation settings as Visual I (total of 114 trials). The third dataset (Opto I) was recorded from
the right V1, evoked by the cortically optical stimulation via a single μ-LED, with an activation

90

voltage of 2.8 V (irradiance of 10 mW/mm^2) for a pulse with a 10 ms duration in a 1 sec trial
(total of 38 trials). All datasets originally were recorded at a sampling frequency of 25 kHz and
down sampled to 1 kHz for data analysis. Each trial was truncated to a 100 ms pre-stimulation
period and an 800 ms post-stimulation period, aligned based on the starting of the stimulus ON.
The Hilbert transform of each trial of ECoG recordings was computed to estimate the
instantaneous phase. The recorded data of each trial was bandpass filtered to extract the signal
around a frequency of interest (theta band: 4 – 8 Hz), then the Hilbert transform was applied to
compute instantaneous phases of the signal. To visualize how instantaneous phases were
modulated by different stimulations, the instantaneous phases of each trial, labeled with different
colors according to Fig. 6-3d, were aligned based on the stimulus ON time and stacked. Both
Visual I (Fig. 6-3a) and Opto I (Fig. 6-3c) datasets demonstrated a clear phase synchronization in
theta band, because the phase values were modulated by the stimulus time. This modulation was
extremely reliable across trials during the stimulation compared with the phases of each trial in
Visual II (Fig. 6-3b). Two observations were made from the Hilbert transform analysis, which
were validated through all channels with a nonparametric test (p < 0.05, Wilcoxon signed rank
test for each pair of channels).


The visual stimulation generated multiple time windows of phase synchronization during the
trial, between 0.2 sec and 0.6 sec. The first phase synchronization event emerged around 100
ms after the stimulation, which is consistent with 100 ms latency observed in human vision
system [180].



The cortically optical stimulation generated almost deterministic phase-locked neuronal
oscillations without any latency for ~ 0.4 sec. The duration of the phase-locked neuronal
oscillation was ~ 400 ms, similar to that of the visual stimulation.

91

Figure 6-4. Broadband (1 – 150 Hz) phase-locked synchrony measured by RID-TFPS: a black
line indicates stimulation ON. Phase-locked synchrony measures from (a) Visual I, (b) Visual II,
and (c) Opto I.
Second, phase synchrony of three datasets in a broadband frequency range was measured
using the RID-TFPS method. The analysis results of selected channels, channel 4 from Visual I
and channel 8 from Opto I, are shown in Fig. 6-4a and c, respectively. High phase-locked
synchrony values in a low frequency band (1 – 25 Hz) in both Visual I and Opto I datasets were
consistent with the phase-locked oscillations observed in the previous analysis based on the
Hilbert transform. Because the RID-TFPS approach can monitor a phase-locked synchrony
throughout the broadband frequency range without bandpass filtering, synchronization in
different frequency bands was detected. For example, for the dataset Visual I, there is increased

92

synchrony (Fig. 6-4a) ~ 0.08 sec after the simulation, spanning multiple bands such as theta band
(4 – 8 Hz) (p < 0.01, Wilcoxon signed rank test for each pair of channels), alpha band (9 – 12
Hz) (p < 2.3e-4, Wilcoxon signed rank test for each pair of channels), and beta band (13 – 30 Hz)
(p < 0.03, Wilcoxon signed rank test for each pair of channels), followed by later
synchronization observed within lower frequency bands such as delta band (0.1 – 3.9 Hz) (p <
2.1e-4, Wilcoxon signed rank test for each pair of channels) and theta band (p < 0.017, Wilcoxon
signed rank test for each pair of channels). Several observations were also made from the
analysis results.


The phase-locked synchrony estimation using RID-TFPS is consistent with the findings in
the previous method based on the Hilbert transform.



Phase-locked neuronal oscillations in multiple bands (theta, alpha, and beta bands from
Visual I and theta, beta and gamma bands from Opto I) were observed.



100 ms latency was observed from Visual I, while immediate phase-locked oscillations were
observed from Opto I.



Stronger and more frequent phase-locked synchronies in gamma band were observed from
Opto I than counterparts from Visual I.

6.7 Conclusion
The influence of cortically optical stimulation on a large population of neurons in rat V1 was
investigated by studying differences between visual stimulation and cortically optical stimulation
evoked neuronal oscillations and their phase-locked synchrony measures. Simultaneous
cortically optical stimulation and light-evoked ECoG recordings were done using a newly
developed Opto-μECoG array. A new time-varying phase synchrony measure, referred to as
reduced interference distribution time-frequency phase synchrony, allows detecting a phase-

93

locked synchrony throughout a large range of frequency bands without bandpass filtering.
Preliminary results revealed reliable and consistent phase-locked synchrony on both the visual
and cortically optical stimulations. Differences in phase modulations between the natural
stimulus driven neural oscillations and the optically stimulated neural oscillations were also
studied, in terms of latency and specific frequency bands of phase-locked synchrony.

94

CHAPTER 7
Chronically implantable wireless hybrid neural
interfaces
7.1 A wireless slanted optrode array
Building on the previous studies, this chapter presents a wireless slanted optrode array with
integrated μ-LEDs, capable of simultaneous light stimulation and electrical neural recording to
achieve a truly untethered bi-directional neural interface. Recording electrodes wrapped around
the waveguide cores permit simultaneous measurement of neural responses from various depths
upon optical excitation. Fig. 7-1 depicts the concept and design of the wirelessly poweredslanted optrode array, containing 32 embedded μ-LED light sources on a 2.5 × 2.5 mm^2
flexible substrate, with 4 × 4 channels per each hemisphere to cover the bilateral visual cortices
in rats. The slanted microneedle array was fabricated separately on a PDMS substrate and
integrated with the multi-LED array using shape-matching assembly. A DBE method, discussed
in Chapter 5 [130] was used to form a slanted microneedle structure, in which the lengths (300 –
1500 μm) of individual microneedles were determined by the height variance along the droplet
curvature. A multilayer metal/Parylene shell was fabricated on the waveguide core to prevent
photoelectrical artifact and improve biocompatibility.
Integrated LED light sources powered by the wireless switched-capacitor stimulator
(SCS) enable a truly untethered system [33]. Inductive power transmission across the skin has
been a viable solution to provide sufficient power to the implantable medical devices (IMDs),
while overcoming size, cost, and longevity constraints of embedded primary batteries [33]. Light

95

sources, however, typically require high instantaneous power to emit sufficient light for optical
neural stimulation, which can be a significant limiting factor in conventional IMDs [181]. Fig. 72a shows the conceptual block diagram of a conventional inductively-powered array of LEDs for
wireless optogenetics, where a rectifier and a regulator convert AC voltage across a secondary
coil, L2, to DC output voltage to supply a LED driver. Power losses in these stages result in poor
overall power efficiency from L2 to the LED. Moreover, high instantaneous power that flows to
the LEDs when they are on, leads to large load variation, which affects the impedance matching
with the inductive link, significantly increasing the required inductive power level, affecting a
safety issue, and degrading the inductive link power efficiency as well as supply voltage for the
rest of the IMD.

Figure 7-1. Concept diagram of wireless optrode array with integrated μ-LEDs

96

LEDs

LED
Driver

L2

Regulator

L1

C2

Rectifier

M

C1

Skin

Power
Tx&Ctrl

High power

(a)

PC
Efficient
charging

Data

High
power
driving

TCON

µ-LED array

L1

C2
L2

GND
Inductive
link

CS

CP
GND

CN

SCS

CM

M

Switch

C1

Charger

VCOIL

Skin

Power Tx &
Data mod.

MCU

Power Control

+
VLED

LED

(b)
Figure 7-2. (a) Simplified block diagram of a conventional inductively-powered device
combined with an array of LEDs for wireless optogenetics and (b) simplified block diagram of
the wireless switched-capacitor stimulating (SCS) system to efficiency drive the μ-LED array.
To address these limitations in implantable optogenetics, Lee et al. has utilized a
switched-capacitor stimulating (SCS) system for power-efficient wireless optical stimulation
[33]. The SCS system efficiently charges a small array of storage capacitors directly from the
inductive link and periodically discharges them into the micro-LED array, providing high
instantaneous current without burdening the inductive link and system supply. To control
stimulation parameters, a custom-designed PC interface wirelessly sends data to the SCS system
through the inductive link, while a commercial neural recording system simultaneously detects
the evoked neural signals from the same 3-D optrode array. Moreover, the proposed hermetic
sealing method and in vitro test results verified the functionality and reliability of the chronic

97

implant prototype, which will be used to prepare the SCS system for implantable optogenetics. In
vivo acute experiments were conducted to demonstrate optical modulation of neural activity with
the integrated LED array powered by the wireless SCS system.

7.2 Methods
7.2.1 Slanted waveguide array
In Chapter 5, a polymer-based fabrication method (DBE) was introduced, utilizing the height
variance in a dome-shaped SU-8 structure to create out-of-plane microneedles with various
lengths, and the slanted optrode array also was designed based on the DBE method.
The optrode enables delivering light into the target area as well as recording neural
activity simultaneously. The basic structure of the optrode comprises a SU-8 microneedle
waveguide core covered by metallic cladding, which is used as a recording electrode and a
shielding layer to prevent light-leakage from the sidewall. However, it is found that if the metal
layer is used for neural recording directly, light-induced artifacts will occur, which require post
processing to recover the original neural data. This phenomenon is consistent with a classical
photoelectrochemical finding, the Becquerel photovoltaic effect. To minimize the light-induced
artifacts, four layers of oxide-polymer-metal-polymer sandwich structure were designed.
Multilayer metal/Parylene shell was fabricated on the outer sidewall of the waveguide core,
which contained an indium tin oxide (ITO) shielding layer to prevent photoelectrical artifact, an
opaque metal layer to block light side-leakage and record neural activity, and Parylene as
insulating layers between ITO and metal and an encapsulation for improved biocompatibility.
Opaque metal was removed from the tip of the waveguide to allow light delivery to the brain
neurons. A profile of the sandwich structure is shown in Fig. 7-4b.

98

7.2.2 Switched-capacitor stimulation (SCS)
Fig. 7-2b shows the simplified block diagram of the wireless switched-capacitor stimulating
(SCS) system for power-efficient optogenetics. Conventional inductively-powered current
stimulators require a rectifier, a regulator, and an array of current sources to generate stimulation
pulses from VCOIL, while power losses at each stage result in poor overall stimulator efficiency
[33]. Contrarily, the proposed SCS system efficiently charges a bank of storage capacitors, CS,
directly from VCOIL through the inductive link without using any rectifiers and regulators,
improving the capacitor charging efficiency [33]. Moreover, the charge stored in capacitors is
delivered to the load through switches, efficiently creating stimulation pulses.
In the inductively-powered SCS system, a power transmitter (Tx) drives the primary coil,
L1, at the power carrier frequency, fC. This signal induces current in the secondary coil, L2,
through the inductive link, generating an AC voltage, VCOIL, across the resonance circuits, L2 and
C2. The SCS system efficiently charges a bank of storage capacitors, CP and CN, directly from
VCOIL through a series capacitor, CS, and an inductive charger without using any rectifiers and
regulators [170], improving capacitor charging efficiency. The charge stored in capacitors is
delivered to the load through switches, efficiently creating stimulation pulses. In addition, a
charge monitoring (CM) circuit measures the amount of injected charge to enable chargebalanced stimulation. For power-efficient wireless optogenetics, the SCS system periodically
discharges the storage capacitors into an array of micro-LEDs to emit sufficient light and evoke
neural activity. After charging, CP and CN pairs are connected in series to provide higher LED
voltage, VLED, for optical stimulation.
Forward data telemetry utilizes the same inductive link to wirelessly set stimulation
parameters in the system, and the on-chip timing controller (TCON) generates timing signals for

99

capacitor charging and LED driving. Hence, the SCS system can be implanted chronically for
experiments with freely behaving animal subjects. The power control unit generates supply
voltages for the rest of the system and prevents the systems from overvoltage due to sudden
strong coupling.

Figure 7-3. 3-D model for in vivo optogenetics experiments with the SCS system. Inset: Optrode
array with micro-LEDs for optical stimulation and penetrating electrodes wrapped around the
waveguides for neural recording.
Fig. 7-3 shows the 3-D model for in vivo wireless optogenetics with the SCS system,
which receives wireless power and data through the inductive link. The SCS ASIC drives the 3D flexible optrode array, which consists of micro-LEDs for optical stimulation, penetrating

100

electrodes for neural recording, and micro-needle waveguides to enable precise and efficient
light delivery to the target tissue with high spatial resolution [32].
(a)

(b)

Figure 7-4. (a) Fabrication process flow for making the slanted optrode array with integrated μLEDs and (b) the profile of the optrode.

101

Neural signals are recorded from penetrating electrodes wrapped around the waveguide core and
only exposed at the tips of the waveguides, through an independent recording setup. In this setup,
the wireless SCS system with slanted optrode arrays enables simultaneous optical stimulation
and electrical neural recording to test an untethered bi-directional neural interface.

7.3 Fabrication
To reduce fabrication complexity, the multi-LED array and the slanted optrode array were
fabricated and calibrated separately. Then the system was assembled by polymer bonding of
individual components with SU-8. Detailed fabrication was divided into two steps: flexible
multi-LED array and slanted optrode array fabrication. Details of flexible multi-LED array
fabrication process was described in Chapter 3.3.1, so, this study only focuses on the fabrication
process of the slanted optrode array (Fig. 7-4a).
(1) A 3” glass wafer was cleaned and put through a dehydration bake, and ~ 50 μm SU-8
layer was spun onto the wafer and patterned as the mock LEDs. (2) A thin layer of PDMS was
spun on the SU-8 master to create cavities matching the shape of the LED. (3) After the PDMS
was cured for 40 min at 95 °C, photoresist (PR) was patterned on the PDMS substrate to expose
7 mm-diameter circles, followed by oxygen plasma treatment to convert the exposed
hydrophobic areas to hydrophilic ones. (4) After removal of the PR mask, ~ 45 μL SU-8 (SU-8
3005) was dispensed on top of the plasma treated PDMS surface using a micropipette and (5)
patterned with the backside exposure to form the microneedles. (6) After SU-8 development, the
array was polished by O2 plasma etching. (7) DC sputtering of a 0.1 μm thick ITO layer was
performed in a Kurt Lesker Axiss PVD System, followed by deposition of 5 μm Parylene-C in a
chemical vapor deposition (CVD) system (PDS2010 Parylene Coater, Specialty Coating System)
and then 1μm thick Au layer in a thermal evaporator (Edward Auto306). Opaque metal was

102

removed using wet etching from the tip of the waveguide to allow light delivery to the brain
neurons and 5 μm Parylene-C was deposited as a protection layer. (8) The Parylene-C film at the
tip of the optrode was removed using reactive-ion etching (RIE), and the membrane was released
from the glass wafer. (9) Finally, the microneedle array with the matched LED cavities was
aligned onto the corresponding LED chips and bonded with polymer adhesive.
(a)

(b)

(c)

(d)

Figure 7-5. Fabricated prototypes of (a) SEM images of a slanted microneedle structure, (b) 32channel optrode array, (c) microscopic image of individual optrode, and (d) optrode array
coupled with μ-LEDs.
Fig. 7-5 shows images of the prototypes of slanted optrode array. SEM image of lengthvarying SU-8 microneedle structure (300 – 800 μm) shown in Fig. 7-5a demonstrated that the
capability of the DBE method for making varying-length microneedles. The electrode-electrolyte
interface impedances of the optrode array were measured at 1 kHz using a built-in electrodeimpedance-testing circuitry in an Intan evaluation board (RHD2132 and RHD2000-EVAL, Intan

103

Tech. LLC, CA, USA). The impedance of the optrode array was ranging from 10 to 500 kΩ,
which is suitable for local field potential recordings. In this study, the impedance of the optrode
was controlled by size of the Parylene opening.

Figure 7-6. Schematic of in vivo animal experiment set-up with the wireless SCS system.

7.4 Results
7.4.1 In vivo animal experiments
In vivo acute animal experiments were performed to demonstrate the ability of the slanted
optrode array to simultaneously modulate and record neural activities in the visual cortex (V1) of
a rat. All procedures were approved by the Institutional Animal Care and Use Committee
(IACUC) at Michigan State University. Prior to the acute experiments, rodent subjects (Sprague-

104

Dawley rats: 250 – 400 g) had neurons in their V1 transfected with ChR2 to enable light
sensitivity.

1. Stimulation frequency
2. LED turn-on time

...

3. Max
LED
voltage
4. Pulse
period

5. Pulse
width

Figure 7-7. SCS system GUI in LabVIEW to wirelessly control the SCS parameters. Inset: LED
driving voltage waveform and its controllable parameters.
As shown in Fig. 7-6, the SCS system was used to power and control the 32-channel LED
array and a 32-channel headstage was used for simultaneous extracellular recording. The SCS
system receives wireless power and data through the inductive link and drives micro-LEDs in the
optrode array. The SCS prototype occupies 3.9 × 3.9 cm^2 on PCB, and includes off-chip
components for testing and optrode connectors. Key parts of this prototype are the SCS chip (5 ×
2.4 mm^2) and 4 pairs of 1 ~ 5 μF off-chip storage capacitors (SMD-0402, 1 × 0.5 mm^2 each).
The SCS system can be shrunk further using a single flex-PCB that can connect directly to the
LED array and include the receiver coil, L2, without connectors or testing components. Storage

105

capacitors also can be placed on the opposite side of the SCS chip, minimizing the PCB size for
implantable optogenetics. Neural signals were recorded simultaneously from the penetrating
electrodes on the same optrode array through an evaluation board and commercial hardwired
recording setup (RHD2132, Intan Technologies, CA, USA).
A graphical user interface (GUI) has been implemented in the LabVIEW environment to
send data packets from the PC to the MCU of the power Tx module, as shown in Fig. 7-7.
Several parameters of the LED driving signal can be adjusted in this setup: 1) stimulation
frequency, 0.5 ~ 2 Hz, 2) LED total turn-on time, 10 ~ 100 ms, 3) LED peak voltage, 2.5 ~ 3.45
V, 4) pulse period, 2 ~ 16 ms, 5) pulse width, 512 μs, 6) positive output connections, 4 channels,
7) negative output connections, fixed, and 8) storage capacitance, 1 ~ 10 μF. While the pulse
train periodically turns the LEDs on and off, the effective turn-on time can be calculated simply
as total turn-on time x (pulse width / pulse period).
Fig. 7-8 shows the LED driving voltage, VLED, for optical stimulation with SCS and lightinduced in vivo LFP results. The LFPs below 500 Hz were recorded using an optrode array with
waveguides in the rat’s brain when the SCS system drove micro-LEDs with a 512 μs pulse train
for 100 ms at 1 Hz and VLED = 2.7 Vpeak and 3.2 Vpeak, as shown in Fig. 7-8a.
A clear light-induced neural activity was observed in LFP (1 – 500 Hz) in time domain
driven by 3.2 Vpeak input voltage (Fig. 7-8b). 3.2 Vpeak input voltage resulting in average
irradiance of 1.4 mW/mm^2 at the tip of the optrode, which is above the minimal irradiance (1
mW/mm^2) required for light-evoked neural response. This author also tried lowering the input
voltage to 2.7 Vpeak (resulting in the average irradiance of 0.35 mW/mm^2), but no significant
neural modulation was observed.

106

1s
100ms

2ms

+

512µs

...

VLED

-

(a)
500

500

LFP w/ VLED=3.2Vpeak

250

Voltage (µV)

Voltage (µV)

LFP w/ VLED=2.7Vpeak

0

-250
-500
0

2

0

-250

Optical Stimulation
(100ms at 1Hz)
1

250

3

4

-500
0

Optical Stimulation
(100ms at 1Hz)
1

2

3

4

Time (s)

Time (s)

(b)
Figure 7-8. Light-induced LEP recording driven by 2.7 Vpeak and 3.2 Vpeak for 100 ms: (a)
LED driving voltage, VLED, for in vivo optogenetics with SCS (b) light-induced local field
potentials (LFP) with VLED = 2.7 Vpeak and 3.2 Vpeak, (c) instantaneous phase with VLED = 2.7
Vpeak (top) and 3.2 Vpeak (bottom) in low frequency band (1-25Hz), and (d) their corresponding
color-coding.

107

Figure 7-8 (cont’d)

-π

-π/2

0

π/2

π

-π -π/2 0 π/2 π
Instantaneous phase and
their corresponding color
coding
(c)

(d)

Clearly to visualize neural oscillations induced by the optrode array, instantaneous phase
of the two datasets (2.7 Vpeak and 3.2 Vpeak input voltages) of light-evoked LFP (1 - 25 Hz)
was measured based on Hilbert transform. The instantaneous phases of each trial (1 s with 100
ms LED on), labeled with different colors according to Fig. 7-8d, were aligned based on the
stimulus ON time and stacked. With 3.2 Vpeak input voltage, light-evoked neural activity shows
clear phase synchronization, while no phase synchrony was observed in neural recording data

108

with 2.7 Vpeak input voltage. This experiment verified the feasibility of wireless optical
stimulation and simultaneous recording via the wireless slanted optrode array.

(1)

(5)

(2)

(6)

(3)

(7)
(4)

Cu Polyimide PR Solder Components Epoxy Parylene-C
Figure 7-9. Fabrication steps of a hermetically-sealed chronic implant on a flexible polyimide
circuit.
7.4.2 Hermetic sealing for chronic implants
Fig. 7-9 shows the fabrication steps of a hermetically-sealed chronic implant on a flexible
polyimide circuit for experiments using freely behaving animals. The flexible polyimide circuit
was fabricated using Pyralux®AP (AP7163E, DuPont) with the following steps: 1) A 3-inch
Pyralux® wafer was cut and cleaned, and a 3-μm thick photoresist (PR) layer was spin coated. 2)
The circuit design was transferred on to the PR using a lithography technique. 3) The circuit was
patterned by wet etching of copper. 4) Through holes were made by a laser cutter. 5) Vias were
made by filling the through holes with solder. 6) Solder paste was applied on the contact pads
and the SCS chip, and other surface mount devices (SMDs) were populated on the pads. The

109

circuit was baked at 200 °C for 5 mins, and extra flux was applied for reflow soldering, if
necessary. 7) Once all components were populated, a thick layer of epoxy (200 ~ 500 μm) was
applied followed by 10-μm Parylene-C coating as a biocompatible package.
To evaluate the performance of epoxy-Parylene-C packages and the possible electrical
failure of the chronic implant, active accelerated-lifetime soak testing was performed in saline at
a higher-than- body temperature. For soak testing, a discrete version of the SCS circuit was
designed using a MCU (MSP430, Texas Instruments, TX, USA), as shown in Fig. 7-10a. This
circuit was programmed to mimic SCS stimulation patterns, once it received wireless power and
data through the inductive link. Dimensions and fabrication process of the discrete circuit were
identical to the SCS system. Four LEDs (LB QH9G, blue 466 nm, OSRAM, Germany) were
integrated on the flexible PCB, and each LED was individually controlled by the MCU. Spring
structured interconnects were utilized to reduce mechanical stress caused by stretching and
twisting movements in the neck and shoulder area of the animal. A MEMS-based receiver (Rx)
coil was fabricated separately to be placed on the back of the animal and connected to the SCS
with flexible wires (392 F high-flex miniature wire, 36 AWG, McMaster-Carr. OH, USA).
Five discrete SCS devices were prepared and immersed in saline at 75 °C, as shown in
Fig. 7-10b. Each was activated by coupling the Rx and Tx coils, and daily the samples were
visually monitored for possible device failure. No delamination or major physical corrosion
occurred after 14 days in accelerated lifetime testing conditions, while the long-term stability
experiment is still ongoing. Using the Arrhenius relationship, preliminary results show that the
implant lifetime can be equivalent to 3.5 years at body temperature of 37 °C, sufficient for oneyear duration requirement in the animal study. Further improvements can be achieved by

110

optimizing temperature and duration of heat treatment for the Parylene/metal thin-film skin or by
using additional chemical treatments.

(a)

(b)

Figure 7-10. (a) Discrete SCS chronic implant and (b) accelerated soak testing in saline with
wireless power and data transfer.
The discrete SCS system on the flexible circuit with the integrated LED array was
chronically implanted to assess the feasibility and stability of the chronic SCS implant. The SCS
board and Rx coil were coated with a thin layer of Polydimethylsiloxane (PDMS) to protect the
surrounding tissue against damages caused by sharp edges. The ChR2 transduced rat was
anesthetized and placed in a stereotaxic apparatus. Using sterile surgical procedures, a 2 ~ 3 cm
incision was made in the skin overlying the skull, and the skin was detached from the skull using
a scalpel. Once bregma and lambda sutures were exposed clearly, a hemostat was inserted under
the skin of dorsal area, detaching the spinotrapezius muscle and skin to create a reservoir for the
discrete SCS implant. The Rx coil was inserted at the center of the dorsal area through the
channel, and the flexible circuit was placed around the rat’s shoulder.

111

After both Rx coil and SCS circuit were in place, the Tx coil was inductively coupled
with the Rx coil through the skin, and the discrete SCS system was tested to verify possible
damage during insertion. Once functionality was confirmed, a small craniotomy was made to
expose V1 using a micro drill (Ideal Micro-Drilltm, Roboz Surgical Instrument Co., MD, USA).
With the LED array in place, dental cement was applied to secure the array in position. A piece
of gel foam was placed on top of the exposed V1 to cover the dental cement. The skin overlying
the skull was sutured closed. After two days of recovery the rat was brought to the lab and
slightly anesthetized with a mixture of oxygen and isoflurane. The discrete SCS implant was
tested daily by visual inspection of the LED light through the skin. The implant was verified for
14 days, while no behavioral changes were observed.

7.5 Conclusion
Design and fabrication process of the wireless slanted optrode array for a really untethered bidirectional neural interface were presented. The array was inductively powered and controlled by
the wireless SCS system, designed to improve power efficiency. The SCS system for implantable
wireless optogenetics provides high instantaneous power through the inductive link to emit
sufficient light and evoke neural activities. The LabVIEW PC interface and custom-designed
power Tx module provide wireless power and data to the SCS system, while electrodes
embedded in the optrode array enable simultaneous neural recording. The self-assembled LED
array on a single substrate can reduce manufacturing cost. Acute in vivo experiments with optical
stimulation and LFP recording have verified the efficacy of the SCS system for wireless
optogenetics.

112

Chapter 8
Behavioral animal experiment
8.1 Motivation
As stated in Chapter 6, high risk is associated with development of new animal model studies,
especially in behaving animal experiments, due to difficulty in interpreting stimuli-driven
behavioral responses in animal models. For this reason, the concept of artificial sensory
perception induced by optical neural stimulation based on optogenetics has been tested in a wellestablished behaving animal experiment to detect artificial auditory sensory perception,
specifically tinnitus in mice model, prior to pursuing proposed behaving animal experiments in
Chapter 6. A new chronically implantable wireless neuromodulator in behavioral mouse models
has been developed to study whether optical modulation of specific subtypes of neural
population can induce chronic tinnitus perception in freely behaving mice.
Chronic tinnitus is a prevalent auditory disorder that affects approximately 5-15% of the
general adult population [182], [183]. An estimated 1-3% of those sufferers experience severe
tinnitus that substantially reduces their quality of life, involving sleep disturbance, work
impairment, and psychiatric distress [184]-[186]. Most cases of chronic tinnitus are associated
with aging and early noise exposure [187]-[190], and at present are incurable [191]. Although
tinnitus initially may be triggered by injuries to the auditory periphery, the neural generator of
tinnitus most likely resides in CNS [192]-[197].
A variety of animal models have been explored in different species, where tinnitus is
normally induced using either ototoxic drugs such as salicylate and quinine or exposure to loud

113

noise, and detected by either measuring animal behavior or imaging abnormal neural activity (in
physiological models) [198], [199]. CBA/CaJ and C57BL/6J mice are well-established mouse
strains to study noise-induced, or age-related hearing loss. They easily are modified by genetic
methods, important for studying molecular pathways underlying hearing loss. This is another
benefit of using a mouse model, since most drawbacks of a rat model in optogenetics study are
the virus injection step. Extensive optogenetic mouse lines are based on these mice and, in the
long run, the transgenic mouse line saves time and money.
Optical neural interface is an ideal solution to current limitations in behavioral mice. This
neural interface should be applicable readily to other neurological disorders associated with
dysfunctional neuronal plasticity. The novel, mass-producible, optogenetics tool for wireless
optical neural modulation, would enable neuroscientists, clinical researchers, and the
pharmaceutical industry to study neuronal mechanisms and behavioral readouts underlying CNS
disorders.

8.2 A simple battery-powered wireless optical neuromodulator for
freely behaving mice
The inductively powered wireless approach, described in the previous chapter, requires internal
implantation of a receiving coil beneath the skin of the animals’ back. The external transmitting
coil must be aligned with the receiving coil that requires an external harness such as a backpack,
and this approach is not feasible for small animals. To overcome this restriction, in addition to
the inductively powered wireless system, this author has developed a simple, lightweight, and
low-cost optical neuromodulator, which combines the flexible LED array with a battery-powered
LED driver circuit, suitable for chronic implantation in freely behaving small animals such as
mice. Depending on requirements of an experiment design, three LED controllers are available: a

114

magnetic switch controller, an infrared (IR) switch controller, and an IR-microcontroller (MCU)
based controller.

Figure 8-1. Basic design of battery-powered wireless optical neuromodulator
Basic design of the proposed battery-powered LEDs array comprises three major parts:
battery holder, LED controller circuit, and LEDs array, as shown in Fig. 8-1. The battery holder
(BS-1225-PC, MPD Inc., NY, USA) serves as a body as well as a power source. The LED
controller circuits were attached on the wall of the battery holder to minimize overall size of the
device, and the custom-designed LED driver was fabricated on a flexible double-layered
polyimide substrate, Pyralux®AP (AP7163E, DuPont, DE, USA), using UV lithography
technique. Operation voltage is between 2.7 V and 3.2 V and it is supplied by a single BR-1225
lithium battery. This 3 V battery has nominal capacity of 48 mAh sufficient to drive two LEDs
for approximately two hours without significant voltage drop. Each LED is connected to a
current limiting resistor in series, and bypass-capacitors are connected to every semiconductor
components. The battery-powered LEDs array can be integrated with a controller circuit or a
flexible wire depending on location of implantation. The overall dimension of the device

115

including the battery is 5 mm × 12 mm × 6 mm and its total weight is less than 1.5 g which is
less than 10% of the subject’s body weight. The entire system comprises fewer than ten off-theshelf components, and the assembly process requires no special equipment. Replaceable batteries
extend the device’s lifetime and eliminate any need for complex circuitry for inductive powering
and/or battery recharging. The optically- transparent Parylene package ensures the device’s
safety and biocompatibility while maintaining its mechanical flexibility.

Figure 8-2. Design of the magnetic-switch based controller and a schematic of the LED driver
circuit
8.2.1 Magnetic-switch based controller
Initially, the magnetic-switch based device was designed for a fully-implantable version of the
device that requires an external switch to activate the implanted device. Basically, a latch circuit
is connected to a capacitive oscillator through a momentary switch, in this case a reed switch.
Once the switch is closed, the oscillator generates a pulse train with a specific frequency
controlled by a capacitor value, and a latch circuit generates the same pulse train. When the
switch is open, the latch switch stores a momentary state of the oscillator and stays. This allows

116

dual activation modes - oscillation mode and continuous activation mode. Attaching a small
magnet on the reed switch activates oscillation with a fixed frequency. Removing the magnet
from the switch triggers continuous activation or turning off the circuit, depending on the
momentary state of the oscillation at the moment of the magnet removal. The latch circuit
consists of a single bipolar junction transistor (BJT), single p-type MOSFET, and two resistors,
and the oscillator comprises a single BJT transistor, two resistors, and one capacitor, thus the
entire circuit requires only eight off-the-shelf components.
The magnetic-switch LED array is suitable for applications requiring a fixed-frequency
modulation. If the experiment requires constant stimulation during an entire session, the system
can be activated by attaching a small magnet on the circuit. Remote activation also is possible by
applying a focused-magnetic field at designated locations, and this method is useful for an
experiment designed to give a stimulus at specific locations and conditions.

Figure 8-3. Design of the IR-switch controller and the schematic of the LED driver circuit
8.2.2 IR-switch based controller
In the IR-switch controller, the LEDs array is controlled wirelessly by a modulated IR signal
generated by an external transmitter system. A digital oscillator with an external trigger is used

117

to generate alternating TTL signals at 38 kHz. The receiver unit is intended to be small to permit
mounting on the skull of a freely-moving mouse. The IR-switch controller circuit comprises a
low-power IR reception module (GP1UD271XK, Sharp, Japan) and an analog switch
(ADG702BRTZ, Analog Devices, USA).
Main advantage of the IR-switch LED array is a variable modulation frequency. LED
output frequency can be controlled externally by changing the TTL trigger of the transmitter
unit. The receiver unit can receive the IR transmitter signal from relatively long distance (< 2 m)
and a group of receiver units can be activated simultaneously.

Figure 8-4. Design of IR-MCU based controller and the schematic of the LED driver circuit
8.2.3 IR-MCU based controller
The aforementioned two devices are capable of activating only a single LED or a group of LEDs
at the same time. That limits spatial resolution of optical stimulation. IR-MCU based LED array
is designed to activate individual LED units independently. A lower-power MCU (MSP430,
Texas Instruments, Dallas, TX) controls activation of four individual LEDs, and the MCU
receives an IR control signal from the external IR transmitter. A low-power IR reception module,
directly connected to an input pin of the MCU, demodulates the signal from the IR transmitter,

118

and the MCU receives a pulse train. Depending upon a switch pressed, the transmitter generates
a modulated signal with a specific number of pulses. Here the MCU works as frequency-counter
as well as band pass filter to detect incoming frequency generated by the transmitter, and it
performs a pre-programmed task based on the number of counts. The proposed IR-MCU based
controller can control each LED independently with different frequencies, and pulse-width
modulation (PWM) in MCU allows not only on-off states of LEDs, but also intensity control of
LED output.

(a)

(b)

(c)

Figure 8-5. Summary of GPIAS for measuring tinnitus (adapted from [183]): (a) an animal is
startled in the presence of a background noise plus a startle pulse. (b) A silent gap 100 ms before
the startle pulse can reliably inhibit the reflex. (c) Due to tinnitus, the gap is not easily detected.

8.3 Behavioral experiment for tinnitus detection in a mouse model
Bao et al. have a long track record in developing genetically-modified animal models to study
behavioral changes in mice. They used neuregulin-1 (NRG1), a molecule important for synaptic
plasticity, to regulate the expression of neuronal neurotransmitter receptors and ion channels in

119

hippocampus [200]-[202]. A reflexive behavioral model, termed the gap-prepulse inhibition of
the acoustic startle (GPIAS), was utilized to detect tinnitus in mice.
8.3.1 GPIAS method
The GPIAS method is based upon the ability of the acoustic startle reflex to be reduced by
a preceding signal/stimulus, in this case a silent gap in an otherwise constant acoustic
background. When an animal effectively detects such a silent gap, its response to a startle
stimulus is reduced in magnitude (Fig. 8-5a and 8-5b). However, the presence of tinnitus would
partially increase this reduction (Fig. 8-5c). The GPIAS method has proven efficient as a rapid
screen for tinnitus in rats; requiring no prior training, no food or water restriction or aversive
shock, and allows for many animals to be tested simultaneously in a relatively short (30 – 40
min) session [203]. Because there is no learned response to maintain, extinction is not an issue
and the measure can be collected repeatedly across the lifespan, lending itself well to
longitudinal studies for this project.
In mice, encouraging data have been obtained by the conventional GPIAS although
additional testing is beneficial to eliminate both false positive and negative animals. The
modified GPIAS method was designed to detect tinnitus induced by optogenetics
neuromodulation. This new method can overcome both false positive and negative detection of
tinnitus by the GPIAS alone, which also effectively overcomes the difficulty of determining the
conscious perception of phantom sound in animals [203]-[205]. Therefore, the combination
method to detect optogenetics stimulation-induced tinnitus in behavioral mice is designed.
8.3.2 The modified GPIAS method
The modified GPIAS method is a two-step test including the conventional GPIAS and the
operational test. To determine whether optogenetics neuromodulation in the auditory cortex

120

induces tinnitus-like behavior, the GPIAS method is first used to detect the presence of optical
stimulation-induced tinnitus among three groups of animals: Normal C57BL/6 control-, Auditory
Cortical “Stimulation-On”-, and Auditory Cortical “Stimulation-Off”- groups. For animals in the
“Stimulation-ON” group, their auditory cortex is stimulated optically with various light
intensities of 1 – 10 mW/mm^2 and durations of 1 – 100 ms, using the implanted optrode
interface. During the GPIAS testing, the animals’ behavioral patterns are monitored continuously
to identify possible presence of tinnitus. As soon as a significant tinnitus-induced behavior is
observed in the “Stimulation-ON” group, the cortical stimulation is turned off to study whether
this light-induced tinnitus effect is reversible. The operational test (shuttle box) is used to further
confirm the data. A conditioned place preference test is used to detect tinnitus-like behavior in
mice. Test details are explained in Chapter 8.4.2.

8.4 Results
8.4.1 Characterization of the battery-powered wireless neuromodulator
Several experiments were conducted to characterize properties of the proposed neuromodulators.
Devices were chronically implanted in a freely behaving mouse and behavioral tests were
performed daily basis to demonstrate long-term stability and capability of the proposed devices.
Since these devices have a designated voltage source, it is important to ensure the array is
capable of evoking neural activity. Irradiances of the battery-powered neural stimulators with
fully charged battery was measured using an optical power meter (Model 815, Newport, CA,
USA). As Fig. 8-6a shows, total irradiances of IR-switch and MCU based devices were in a
similar range (4 – 6 mW/mm^2), and these values were well above the minimum requirement for
activation of ChR2 (1 – 5 mW/mm^2). Although the magnetic-switch based device produced
less optical output power than did others, that power was sufficient to evoke neural activity of

121

the ChR2 expressed neurons.

(a)
(b)
Figure 8-6. Optical characterizations of the battery-powered wireless optical neuromodulator.
The capacity of a single battery to maintain a certain light intensity (1 mW/mm^2) was also
measured. The IR-switch stimulator was activated at 50 Hz with duration of 10 ms for 5 hours
continuously, and light intensity was measured every 10 mins. Slow exponential decaying of the
LED light intensity was observed as shown in Fig. 8-6b. These results support the necessity of
the replaceable battery for a long-term behaving animal experiment.

Figure 8-7. Chronic implantation of the LED controllers in freely-moving mice.

122

8.4.2 In vivo animal experiment
To evaluate safety and long-term reliability of the battery-powered wireless stimulator prior to
the behaving animal experiment, devices were chronically implanted on the skulls of six mice (2
devices of each type). Details of the surgical procedure of the proposed in vivo chronic implant
animal experiments are explained.
All procedures were approved by the Institutional Animal Care and Use Committee
(IACUC) at Michigan State University and Northeast Ohio Medical University. Transgenic mice
(B6.Cg-Tg(Thy1-COP4/EYFP)18Gfng/J) were used in this study. These transgenic mice express
ChR2 fused to enhanced-yellow fluorescent protein (EYFP) under the control of the mouse
thymus cell antigen 1 (Thy1) promoter. Mice underwent surgery to expose the dorsal surface of
the auditory cortex. Prior to surgery, they were anesthetized with 4% isoflurane, then transferred
to a small animal stereotaxic device. The hair over the dorsal surface of the skulls was removed
(either clipped with scissors or by depilatory agent) and the skin was cleaned (alcohol and
Betadine). Skulls then were exposed by a midline incision in the skin, the overlying muscles
were reflected laterally and a small hole (<100 µm diameter) made in the skull over the region of
the auditory cortex. Incision length was approximately 20 mm, the minimum needed to attach the
wireless device. The LED array was placed through the hole and into the auditory cortex, fixed
to the overlying skull and cemented in place, closing the hole. Next, the battery holder was
placed on top of the LED array and cemented to the skull using dental cement (3M ESPE Ketac
Cem radiopaque, 3M, MN, USA). Topical lidocaine was applied to alleviate pain from surgical
wounds. After surgery, the animals were placed in their cages to recover from the anesthetic.
Once the animals awoke, they were housed back in the animal facility and allowed to recover for
three days prior to behavioral testing.

123

After recovery the mice were brought to the lab and the implanted devices tested daily by
activating the LEDs for 1 hour (Fig. 8-7). For more than two months, none of the six chronically
implanted devices failed and the result demonstrates that the proposed battery-powered wireless
neurostimulator is suitable for a long-term behaving animal experiment.
8.4.3 Behavioral animal experiment
The GPIAS test is similar to the previously-published method [183]. In short, the test was
conducted with Kinder Scientific's GPIAS hardware and software (Kinder Scientific, CA, USA).
Gap detection and prepulse inhibition (PPI) testing used background sounds presented through
one speaker (Vifa XT25TG30-04) and startle stimuli presented through a second speaker
(Powerline CTS KSN-1005) located in the ceiling of the testing chamber, 15 cm above the
animal's head. The floor of the chamber was attached to a piezo transducer and provided a
measure of startle force applied to the floor. A clear polycarbonate animal holder, with holes cut
for sound passage, was suspended above the floor allowing the animal to freely turn around
while minimizing excessive movement. The startle-eliciting stimulus was a 115-dB SPL, 20-ms
duration broadband noise with a rapid rise gate (<0.1-ms). PPI trials were periodically measured
as a control condition. Gap detection trials are identical to PPI trials except for the inserted gap in
the otherwise continuous background, as opposed to the inserted prepulse stimulus in the
otherwise quiet background. Data from gap detection or PPI trials yield a single value expressed
as a ratio of the mean response in gap trials/ mean response in startle-only trials. A response of
1.0 would mean that the startle reflex is the same whether or not preceded by a gap or prepulse
stimulus (depending on the trial used). The lower the ratio is from 1.0, the better the startle
inhibition. Individual and group data were first inspected for outliers, distributions, and
descriptive statistics. Data were then subjected to analysis of variance (ANOVA) hypothesis

124

tests. All statistical analyses were conducted using Prism Graph Pad 5.0c (Graph Pad, CA,
USA).
A conditioned place preference test was used to detect tinnitus-like behavior in mice.
First, in a shuttle box, mice were trained to move into the illuminated compartment to avoid foot
shocks when a sound is present, and stay in the dark side when it is silent. Among these trials, a
total of 12 probe trials was randomly introduced in which no sound was played and no foot
shock was delivered. Then, after the noise exposure, mice underwent the same test. Development
of tinnitus-like behavior was detected based on mice preference for the illuminated compartment
in the probe trials.

8.5 Conclusion
As a part of development of a cortically visual prosthesis based on optogenetics, the concept of
artificial sensory perception induced by optical neural stimulation was tested in the GPIAS test
for tinnitus detection in mice model. Three types of chronic wireless neuromodulators were
developed to adapt the hybrid optical neural interfaces to a chronically implantable device on
behaving small animals, specifically in a mouse model. Each device was designed to serve
different applications, and the in vivo chronic animal experiment result proved that the proposed
battery-powered wireless neuromodulators are suitable for long-term freely behaving animal
experiment. The modified GPIAS test was proposed to overcome the limitation of the
conventional GPIAS test alone for tinnitus detection in mice model, and currently the evaluation
of the proposed method is in progress.

125

Chapter 9
Conclusion
Electrical neural stimulation has been a powerful tool for decades in electrophysiology and
clinical neuroscience. However, its limitations necessitate improved technology. Recent
developments in optogenetics have demonstrated the ability to target specific types of neurons
with sub-millisecond temporal precision via direct optical stimulation of genetically modified
neurons in the brain. Optogenetics has become a popular technique that has not only replaced
electrical neuromodulation, but also opened a new field in the study of the nervous systems. The
success of optogenetics has also opened new research area including the development of new
light-sensitive proteins and technologies for light delivery.
In this dissertation, a series of hybrid neural interfaces, combining optogenetics-based
optical neuromodulation and electrical neural recordings, have been designed, fabricated, and
characterized

to

demonstrate

capability

of

simultaneous

optical

stimulation

and

electrophysiological recording of cortical neurons. The proposed hybrid neural interfaces allow
the possibility of achieving a truly untethered system with integrated multi-μLEDs array crucial
for chronic implants. 3-D light distribution using the 3-D waveguide array significantly improves
the spatial resolution of light delivery.
The development of the hybrid neural interfaces has led an invention of several enabling
technologies. Self-assembly of μ-LEDs using a PDMS stamp and low melting point solder
enables a large population of micro-scale chip assembly on flexible polymer substrate possible.
The DBE method for making slanted microneedles was designed, developed, and characterized.
The results have shown that the three key parameters of the microneedle (the length, tip and base
126

diameters) can be controlled precisely by manipulating the droplet volume and the mask design
(the aperture diameter and the distance from the center). The measurements of these parameters
were consistent with the analytical estimation, demonstrating the repeatability and controllability
of the DBE method for fabricating varying-length microneedles.
Based on works presented above, this author aims to develop a cortically based optoepidural visual prosthesis for therapeutic management of all forms of blindness. The hypothesis
that epidural optical stimulation of a large population of ChR2 expressing neurons in the primary
visual cortex (V1) can induce phosphenes was formulated based on the studies of electricallybased cortical visual prosthesis. As a proof-of-concept study, three major tasks have been
identified including stimulus-driven ECoG data analysis, a chronic device development for freely
behaving animal experiment, and a behavioral animal test to examine the ability of optical
stimulation of ChR2 expressing neurons to induce phosphenes.
First, stimulus-driven ECoG data have been analyzed to study the influence of cortically
optical stimulation on a large population of neurons in rat V1. Preliminary results demonstrated
reliable and consistent phase-locked synchrony on both the visual and cortically optical
stimulations. Differences in phase modulations between the natural stimulus driven neural
oscillations and the optically stimulated neural oscillations were also studied, in terms of latency
and specific frequency bands of phase-locked synchrony. More stimulation parameters, such as
duration, light intensity, and size of stimulation site, must be explored to improve the OptoμECoG system, to replace a natural sensory feedback system.
Second, a chronic device for freely behaving animal experiment has been developed based on
inductive link-based powering and data transfer technology. The wireless slanted optrode array
for a really untethered bi-directional neural interface has been developed, and the array was

127

inductively powered and controlled by the wireless SCS system. Acute in vivo experiments with
optical stimulation and LFP recording have verified the efficacy of the SCS system for wireless
optogenetics. Moreover, the hermetic sealing method for chronic implantation is under
development to enable implantable optogenetics with the SCS system.
Finally, the concept of artificial sensory perception induced by optical neural stimulation
based on optogenetics has been tested in a well-established behavioral animal experiment to
detect artificial auditory sensory perception prior to pursuing proposed behaving animal
experiments. A new chronically implantable wireless neuromodulator in behavioral mouse
models has been developed to induce chronic tinnitus perception in freely behaving mice.

128

BIBLIOGRAPHY

129

BIBLIOGRAPHY

[1]

M. A. Lebedev and M. A. L. Nicolelis, “Brain–machine interfaces: past, present and
future,” Trends in Neurosciences, vol. 29, no. 9, pp. 536–546, 2006.

[2]

Á. Barbero and M. Grosse-Wentrup, “Biased feedback in brain-computer interfaces,”
Journal of neuroengineering and rehabilitation, vol. 7, no. 34, 2010.

[3]

B. Blankertz, R. Tomioka, S. Lemm, M. Kawanabe, and K.-R. Muller, “Optimizing
spatial filters for robust EEG single-trial analysis,” Signal Processing Magazine, IEEE,
vol. 25, no. 1, pp. 41–56, 2008.

[4]

A. Abbott, “Neuroprosthetics: In search of the sixth sense,” Nature, vol. 442, no. 7099,
pp. 125–127, 2006.

[5]

T. Hinterberger, N. Neumann, M. Pham, A. K bler, A. Grether, N. Hofmayer, B.
Wilhelm, H. Flor, and N. Birbaumer, “A multimodal brain-based feedback and
communication system,” Experimental Brain Research, vol. 154, no. 4, pp. 521–526,
2004.

[6]

L. R. Hochberg, D. Bacher, B. Jarosiewicz, N. Y. Masse, J. D. Simeral, J. Vogel, S.
Haddadin, J. Liu, S. S. Cash, P. van der Smagt, and J. P. Donoghue, “Reach and grasp
by people with tetraplegia using a neurally controlled robotic arm,” Nature, vol. 485,
no. 7398, pp. 372–375, 2013.

[7]

D. M. Taylor, “Direct Cortical Control of 3D Neuroprosthetic Devices,” Science, vol.
296, no. 5574, pp. 1829–1832, 2002.

[8]

J. Carmena, “Becoming Bionic,” Spectrum, vol. 49, no. 3, pp. 24-29, 2012.

[9]

K. Shindo, K. Kawashima, J. Ushiba, N. Ota, M. Ito, T. Ota, A. Kimura, and M. Liu,
“Effects of Neurofeedback Training with an Electroencephalogram-Based
BrainComputer Interface for Hand Paralysis in Patients with Chronic Stroke: A
Preliminary Case Series Study,” Journal of Neurotherapy, vol.14, no.3, pp.229-242,
2011.

[10]

N. Birbaumer, A. R. Murguialday, and L. Cohen, “Brain–computer interface in
paralysis,” Current Opinion in Neurology, vol. 21, no. 6, pp. 634–638, 2008.

[11]

P. S. Motta and J. W. Judy, “Multielectrode microprobes for deep-brain stimulation
fabricated with a customizable 3-D electroplating process,” Biomedical Engineering,
IEEE Transactions on, vol. 52, no. 5, pp.923-933, 2005.

130

[12]

X. Navarro, T. B. Krueger, N. Lago, S. Micera, T. Stieglitz, and P. Dario, “A critical
review of interfaces with the peripheral nervous system for the control of
neuroprostheses and hybrid bionic systems,” Journal of the Peripheral Nervous
System, vol. 10, no. 3, pp. 229–258, Sep. 2005.

[13]

Y. Yao, M. N. Gulari, J. A. Wiler, and K. D. Wise, “A microassembled low-profile
three-dimensional microelectrode array for neural prosthesis applications,” Journal of
Microelectromechanical Systems, vol. 16, no. 4, pp. 977–988, 2007.

[14]

K. Kwon, “Neuroquest: A comprehensive tool for large-scale neural data analysis,”
Michigan State University, 2011.

[15]

H. S. Mayberg, A. M. Lozano, V. Voon, H. E. McNeely, D. Seminowicz, C. Hamani,
J. M. Schwalb, and S. H. Kennedy, “Deep Brain Stimulation for Treatment-Resistant
Depression,” Neuron, vol. 45, no. 5, pp. 651–660, 2005.

[16]

R. E. Kass, V. Ventura, and E. N. Brown, “Statistical issues in the analysis of neuronal
data,” Journal of Neurophysiology, vol. 94, no. 1, pp. 8–25, 2005.

[17]

E. N. Brown, R. E. Kass, and P. P. Mitra, “Multiple neural spike train data analysis:
state-of-the-art and future challenges,” Nature Neuroscience, vol. 7, no. 5, pp. 456–
461, 2004.

[18]

E. C. Leuthardt, G. Schalk, J. R. Wolpaw, J. G. Ojemann, and D. W. Moran, “A brain–
computer interface using electrocorticographic signals in humans,” Journal of Neural
Engineering, vol. 1, no. 2, p. 63, 2004.

[19]

J. Viventi, D.-H. Kim, L. Vigeland, E. S. Frechette, J. A. Blanco, Y.-S. Kim, A. E.
Avrin, V. R. Tiruvadi, S.-W. Hwang, A. C. Vanleer, D. F. Wulsin, K. Davis, C. E.
Gelber, L. Palmer, J. Van der Spiegel, J. Wu, J. Xiao, Y. Huang, D. Contreras, J. A.
Rogers, and B. Litt, “Flexible, foldable, actively multiplexed, high-density electrode
array for mapping brain activity in vivo,” Nature Neuroscience, vol. 14, no. 12, pp.
1599–1605, 2011.

[20]

G. S. Brindley and W. S. Lewin, “The sensations produced by electrical stimulation of
the visual cortex,” The Journal of physiology, vol. 196, pp. 479-493, 1968.

[21]

V. Gilja, C. A. Chestek, I. Diester, J. M. Henderson, K. Deisseroth, and K. V. Shenoy,
“Challenges and Opportunities for Next-Generation Intracortically Based Neural
Prostheses,” Biomedical engineering, IEEE Transactions on, vol. 58, no. 7, pp. 1891–
1899, 2012.

[22]

F. Zhang, A. Aravanis, A. Adamantidis, L. de Lecea, and K. Deisseroth, “Circuitbreakers: optical technologies for probing neural signals and systems,” Nature Reviews
Neuroscience, vol. 8, no. 8, pp. 577–581, 2007.

[23]

F. Zhang, L.-P. Wang, M. Brauner, J. F. Liewald, K. Kay, N. Watzke, P. G. Wood, E.
Bamberg, G. Nagel, and A. Gottschalk, “Multimodal fast optical interrogation of

131

neural circuitry,” Nature, vol. 446, no. 7136, pp. 633–639, 2007.
[24]

A. Zorzos and E. Boyden, “Multiwaveguide implantable probe for light delivery to sets
of distributed brain targets,” Optics letters, 2010.

[25]

T. Abaya, M. Diwekar, S. Blair, P. Tathireddy, L. Rieth, G. A. Clark, and F.
Solzbacher, “Characterization of a 3D optrode array for infrared neural stimulation,”
Biomedical Optics Express, vol. 3, no. 9, pp. 2200–2219, 2012.

[26]

B. McGovern, R. B. Palmini, N. Grossman, E. Drakakis, V. Poher, M. Neil, and P.
Degenaar, “A New Individually Addressable Micro-LED Array for Photogenetic
Neural Stimulation,” Biomedical Circuits and Systems, IEEE Transactions on, vol. 4,
no. 6, pp. 469–476, 2010.

[27]

R. Hishida, K. Watanabe, and E. al, “Transcranial electrical stimulation of corticocortical connections in anesthetized mice,” Journal of Neuroscience, no. 201, pp. 315321, 2011.

[28]

Y. Iwai, S. Honda, H. Ozeki, and E. al, “A simple head-mountable LED device for
chronic stimulation of optogenetic molecules in freely moving mice,” Neuroscience
research, no. 70, pp. 124-127, 2011.

[29]

K. Kwon, B. Sirowatka, A. J. Weber, and W. Li, “Opto-μECoG Array: A Hybrid
Neural Interface with Transparent μECoG Electrode Array and Integrated LEDs for
Optogenetics,” IEEE Transactions on Biomedical Circuits and Systems, vol. 7, no. 5,
pp. 593–600, 2013.

[30]

K. Kwon and W. Li, “Integrated multi-LED array with three-dimensional polymer
waveguide for optogenetics,” Micro Electro Mechanical Systems (MEMS), 2013 IEEE
26th International Conference on, pp. 1017–1020, 2013.

[31]

K. Kwon, A. Khomenko, M. Haq, and L. Wen, "Integrated slanted microneedle-LED
array for optogenetics," Engineering in Medicine and Biology Society (EMBC), 2013
Annual International Conference of the IEEE, pp. 249-252, 2013.

[32]

K. Kwon, H.-M. Lee, M. Ghovanloo, A. Weber, and W. Li, “A wireless slanted
optrode array with integrated micro leds for optogenetics,” Micro Electro Mechanical
Systems (MEMS), 2014 IEEE 27th International Conference on, pp. 813–816, 2014.

[33]

H. M. Lee, H. Park, H. Park, M. Ghovanloo, and M. Ghovanloo, “A power-efficient
wireless system with adaptive supply control for deep brain stimulation,” IEEE
Journal of Solid-State Circuits, vol. 48, no. 9, pp. 2203–2216, 2013.

[34]

K. Kwon, S. Aviyente, and W. Li, “Investigation of phase-locked neuronal oscillation
with optical stimulation based on a time-frequency approach,” Neural Engineering
(NER), 2013 6th International IEEE/EMBS Conference on, pp. 423–426, 2013.

[35]

O. Yizhar, L. E. Fenno, T. J. Davidson, M. Mogri, and K. Deisseroth, “Optogenetics in

132

Neural Systems,” Neuron, vol. 71, no. 1, pp. 9–34, 2011.
[36]

E. S. Boyden, F. Zhang, E. Bamberg, G. Nagel, and K. Deisseroth, “Millisecondtimescale, genetically targeted optical control of neural activity,” Nature Neuroscience,
vol. 8, no. 9, pp. 1263–1268, 2005.

[37]

X. Liu and S. Tonegawa, “Optogenetics 3.0,” Cell, vol. 141, pp. 22-24, 2010.

[38]

B. Y. Chow and E. S. Boyden, “Optogenetics and Translational Medicine,” Science
Translational Medicine, vol. 5, no. 177, pp. 1–4, 2013.

[39]

L. Fenno, O. Yizhar, and K. Deisseroth, “The development and application of
optogenetics,” Annual review of Neuroscience, vol. 34, pp. 389–412, 2011.

[40]

A. Berndt, P. Schoenenberger, J. Mattis, K. M. Tye, K. Deisseroth, P. Hegemann, and
T. G. Oertner, “High-efficiency channelrhodopsins for fast neuronal stimulation at low
light levels,” Proceedings of the National Academy of Sciences of the United States of
America, vol. 108, no. 18, pp. 7595–7600, 2011.

[41]

S. Kleinlogel, K. Feldbauer, R. E. Dempski, H. Fotis, P. G. Wood, C. Bamann, and E.
Bamberg, “Ultra light-sensitive and fast neuronal activation with the Ca2+-permeable
channelrhodopsin CatCh,” Nature Neuroscience, vol. 14, no. 4, pp. 513–518, 2011.

[42]

L. A. Gunaydin, O. Yizhar, A. Berndt, V. S. Sohal, K. Deisseroth, and P. Hegemann,
“Ultrafast optogenetic control,” Nature Neuroscience, vol. 13, no. 3, pp. 387–392,
2010.

[43]

J. Y. Lin, M. Z. Lin, P. Steinbach, and R. Y. Tsien, “Characterization of Engineered
Channelrhodopsin Variants with Improved Properties and Kinetics,” Biophysical
Journal, vol. 96, no. 5, pp. 1803–1814, 2009.

[44]

H. Wang, Y. Sugiyama, T. Hikima, and E. Sugano, “Molecular determinants
differentiating photocurrent properties of two channelrhodopsins from
chlamydomonas,” Journal of Biological Chemistry, vol.284, no.9, pp.5685–5696,
2009.

[45]

V. Gradinaru, K. Thompson, F. Zhang, M. Mogri, K. Kay, M. Schneider, and K.
Deisseroth, “Targeting and readout strategies for fast optical neural control in vitro and
in vivo,” Journal of Neuroscience, vol. 27, no. 52, pp. 14231-14238, 2007.

[46]

G. Nagel, M. Brauner, J. F. Liewald, and N. Adeishvili, “Light Activation of
Channelrhodopsin-2 in Excitable Cells of Caenorhabditis elegans Triggers Rapid
Behavioral Responses,” Current Biology, vol. 15, pp. 2279-2284, 2005.

[47]

M. Prigge, F. Schneider, and S. P. Tsunoda, “Color-tuned channelrhodopsins for
multiwavelength optogenetics,” J. Biol. Chem., vol. 287, no. 38, pp. 31804–31812,
2012.

133

[48]

J. Mattis, K. M. Tye, E. A. Ferenczi, C. Ramakrishnan, D. J. O'Shea, R. Prakash, L. A.
Gunaydin, M. Hyun, L. E. Fenno, V. Gradinaru, O. Yizhar, and K. Deisseroth,
“Principles for applying optogenetic tools derived from direct comparative analysis of
microbial opsins,” Nature Methods, vol. 9, no. 2, pp. 159–172, 2011.

[49]

L. Wen, H. Wang, S. Tanimoto, R. Egawa, Y. Matsuzaka, H. Mushiake, T. Ishizuka,
and H. Yawo, “Opto-Current-Clamp Actuation of Cortical Neurons Using a
Strategically Designed Channelrhodopsin,” PLoS One, vol. 5, no. 9, p. e12893, 2010.

[50]

B. J. Tromberg, N. Shah, R. Lanning, and A. Cerussi, “Non-Invasive In Vivo
Characterization of Breast Tumors Using Photon Migration Spectroscopy,” Neoplasia,
vol. 2, no. 1-2, pp. 26-40, 2000.

[51]

F. Zhang, M. Prigge, F. Beyrière, S. P. Tsunoda, J. Mattis, O. Yizhar, P. Hegemann,
and K. Deisseroth, “Red-shifted optogenetic excitation: a tool for fast neural control
derived from Volvox carteri,” Nature Neuroscience, vol. 11, no. 6, pp. 631–633, 2008.

[52]

O. Yizhar, L. E. Fenno, M. Prigge, F. Schneider, T. J. Davidson, D. J. O'Shea, V. S.
Sohal, I. Goshen, J. Finkelstein, J. T. Paz, K. Stehfest, R. Fudim, C. Ramakrishnan, J.
R. Huguenard, P. Hegemann, and K. Deisseroth, “Neocortical excitation/inhibition
balance in information processing and social dysfunction,” Nature, vol. 477, no. 7363,
pp. 171–178, 2011.

[53]

J. Y. Lin, P. M. Knutsen, A. Muller, D. Kleinfeld, and R. Y. Tsien, “ReaChR: a redshifted variant of channelrhodopsin enables deep transcranial optogenetic excitation,”
Nature Neuroscience, vol. 16, no. 10, pp. 1499–1508, 2013.

[54]

X. Han and E. S. Boyden, “Multiple-Color Optical Activation, Silencing, and
Desynchronization of Neural Activity, with Single-Spike Temporal Resolution,” PLoS
One, vol. 2, no. 3, p. e299, 2007.

[55]

S. Zhao, C. Cunha, F. Zhang, Q. Liu, B. Gloss, K. Deisseroth, G. J. Augustine, and G.
Feng, “Improved expression of halorhodopsin for light-induced silencing of neuronal
activity,” Brain Cell Bio, vol. 36, no. 1, pp. 141–154, 2008.

[56]

V. Gradinaru, K. R. Thompson, and K. Deisseroth, “eNpHR: a Natronomonas
halorhodopsin enhanced for optogenetic applications,” Brain Cell Bio, vol. 36, no. 1,
pp. 129–139, 2008.

[57]

B. Y. Chow, X. Han, A. S. Dobry, X. Qian, A. S. Chuong, and M. Li, “Highperformance genetically targetable optical neural silencing by light-driven proton
pumps,” Nature, vol. 463, no. 7277, pp. 98-102, 2010.

[58]

A. S. Chuong, M. L. Miri, V. Busskamp, and G. Matthews, “Noninvasive optical
inhibition with a red-shifted microbial rhodopsin,” Nature Neuroscience, pp.1-11
2014.

[59]

J. A. Cardin, M. C. E. n, K. Meletis, U. Knoblich, F. Zhang, K. Deisseroth, L.-H. Tsai,

134

and C. I. Moore, “Targeted optogenetic stimulation and recording of neurons in vivo
using cell-type-specific expression of Channelrhodopsin-2,” Nature Protocol, vol. 5,
no. 2, pp. 247–254, 2010.
[60]

A. Berndt, O. Yizhar, L. A. Gunaydin, P. Hegemann, and K. Deisseroth, “Bi-stable
neural state switches,” Nature Neuroscience, vol. 12, no. 2, pp. 229–234, 2008.

[61]

C. Bamann, R. Gueta, S. Kleinlogel, G. Nagel, and E. Bamberg, “Structural Guidance
of the Photocycle of Channelrhodopsin-2 by an Interhelical Hydrogen Bond,”
Biochemistry, vol. 49, no. 2, pp. 267–278, 2010.

[62]

R. D. Airan, K. R. Thompson, L. E. Fenno, H. Bernstein, and K. Deisseroth,
“Temporally precise in vivo control of intracellular signaling,” Nature, vol. 458, pp.
1025–1029, 2009.

[63]

E. Oh, T. Maejima, C. Liu, E. Deneris, and S. Herlitze, “Substitution of 5-HT1A
Receptor Signaling by a Light-activated G Protein-coupled Receptor,” Journal of
Biological Chemistry, vol. 285, no. 40, pp. 30825–30836, 2010.

[64]

M. Stierl, P. Stumpf, D. Udwari, R. Gueta, R. Hagedorn, A. Losi, W. Gartner, L.
Petereit, M. Efetova, M. Schwarzel, T. G. Oertner, G. Nagel, and P. Hegemann, “Light
Modulation of Cellular cAMP by a Small Bacterial Photoactivated Adenylyl Cyclase,
bPAC, of the Soil Bacterium Beggiatoa,” Journal of Biological Chemistry, vol. 286,
no. 2, pp. 1181–1188, 2011.

[65]

M. J. Kennedy, R. M. Hughes, L. A. Peteya, J. W. Schwartz, M. D. Ehlers, and C. L.
Tucker, “Rapid blue-light-mediated induction of protein interactions in living cells,”
Nat. Methods, vol. 7, no. 12, pp. 973–975, 2010.

[66]

S. Konermann, M. D. Brigham, A. Trevino, and P. D. Hsu, “Optical control of
mammalian endogenous transcription and epigenetic states,” Nature,
doi:10.1038/nature12466, 2013.

[67]

B. E. Losavio, V. Iyer, and S. Patel, “Acousto-optic laser scanning for multi-site photostimulation of single neurons in vitro,” Journal of Neural Engineering, pp. 1–13, 2010.

[68]

P. Zhu, O. Fajardo, J. Shum, Y.-P. Z. Scharer, and R. W. Friedrich, “High-resolution
optical control of spatiotemporal neuronal activity patterns in zebrafish using a digital
micromirror device,” Nature Protocol, vol. 7, no. 7, pp. 1410–1425, 2012.

[69]

A. B. Arrenberg, D. Y. R. Stainier, H. Baier, and J. Huisken, “Optogenetic Control of
Cardiac Function,” Science, vol. 330, no. 6006, pp. 971–974, 2010.

[70]

I. Reutsky-Gefen, L. Golan, N. Farah, and A. Schejter, “Holographic optogenetic
stimulation of patterned neuronal activity for vision restoration,” Nature, vol. 4, p.
1509, 2013.

[71]

F. Zhang, V. Gradinaru, A. R. Adamantidis, R. Durand, R. D. Airan, L. De Lecea, and

135

K. Deisseroth, “Optogenetic interrogation of neural circuits: technology for probing
mammalian brain structures,” Nature Protocol, vol. 5, no. 3, pp. 439–456, 2010.
[72]

A. M. Aravanis, L.-P. Wang, F. Zhang, L. A. Meltzer, M. Z. Mogri, M. B. Schneider,
and K. Deisseroth, “An optical neural interface: in vivo control of rodent motor cortex
with integrated fiberoptic and optogenetic technology,” Journal of Neural
Engineering, vol. 4, no. 3, pp. S143–S156, 2007.

[73]

[X. Han, X. Qian, J. G. Bernstein, H.-H. Zhou, G. T. Franzesi, P. Stern, R. T. Bronson,
A. M. Graybiel, R. Desimone, and E. S. Boyden, “Millisecond-Timescale Optical
Control of Neural Dynamics in the Nonhuman Primate Brain,” Neuron, vol. 62, no. 2,
pp. 191–198, 2009.

[74]

S. Royer, B. V. Zemelman, M. Barbic, A. Losonczy, G. Buzsáki, and J. C. Magee,
“Multi-array silicon probes with integrated optical fibers: light-assisted perturbation
and recording of local neural circuits in the behaving animal,” European Journal of
Neuroscience, vol. 31, no. 12, pp. 2279–2291, 2010.

[75]

E. Stark, T. Koos, and G. Buzsáki, “Diode probes for spatiotemporal optical control of
multiple neurons in freely moving animals,” Journal of Neurophysiology, vol. 108, no.
1, pp. 349–363, 2012.

[76]

T. I. Kim, J. G. McCall, Y. H. Jung, X. Huang, E. R. Siuda, Y. Li, J. Song, Y. M.
Song, H. A. Pao, R. H. Kim, C. Lu, S. D. Lee, I. S. Song, G. Shin, R. Al-Hasani, S.
Kim, M. P. Tan, Y. Huang, F. G. Omenetto, J. A. Rogers, and M. R. Bruchas,
“Injectable, Cellular-Scale Optoelectronics with Applications for Wireless
Optogenetics,” Science, vol. 340, no. 6129, pp. 211–216, 2013.

[77]

A. R. Adamantidis, F. Zhang, A. M. Aravanis, K. Deisseroth, and L. de Lecea, “Neural
substrates of awakening probed with optogenetic control of hypocretin neurons,”
Nature, vol. 450, no. 7168, pp. 420–424, 2007.

[78]

S. K. Mohanty and N. V. Thakor “Optogenetics: Optical Methods for Cellular
Control,” Proceedings of SPIE, vol. 8586, 2013.

[79]

A. Zorzos, J. Scholvin, E. Boyden, and C. Fonstad, “Three-dimensional
multiwaveguide probe array for light delivery to distributed brain circuits,” Optics
Letters, vol. 37, no. 23, pp. 4841–4843, 2012.

[80]

R. Pashaie and R. R. Falk, “Single optical fiber probe for fluorescence detection and
optogenetic stimulation,” IEEE Transactions on Biomedical Engineering, vol. 60, no.
2, pp. 268–280, 2013.

[81]

F. Pisanello, L. Sileo, I. A. Oldenburg, M. Pisanello, L. Martiradonna, J. A. Assad, B.
L. Sabatini, and M. De Vittorio, “Multipoint-Emitting Optical Fibers for Spatially
Addressable In Vivo Optogenetics,” Neuron, vol. 82, pp. 1–10, 2014.

[82]

Z. V. Guo, A. C. Hart, and S. Ramanathan, “Optical interrogation of neural circuits in

136

Caenorhabditis elegans,” Nature Methods, vol. 6, no. 12, pp. 891–896, 2009.
[83]

V. Nikolenko, B. O. Watson, and R. Araya, “SLM microscopy: scanless two-photon
imaging and photostimulation with spatial light modulators,” Frontiers in neural
circuits, vol. 2, no. 5, pp.1-14, 2008.

[84]

M. Zahid, M. Vélez-Fort, E. Papagiakoumou, C. Ventalon, M. C. Angulo, and V.
Emiliani, “Holographic Photolysis for Multiple Cell Stimulation in Mouse
Hippocampal Slices,” PLoS One, vol. 5, no. 2, p. e9431, 2010.

[85]

E. Papagiakoumou, F. Anselmi, A. Bègue, V. de Sars, J. Glückstad, E. Y. Isacoff, and
V. Emiliani, “Scanless two-photon excitation of channelrhodopsin-2,” Nature
Methods, vol. 7, no. 10, pp. 848–854, 2010.

[86]

J. Zhang, F. Laiwalla, J. A. Kim, H. Urabe, R. van Wagenen, Y.-K. Song, B. W.
Connors, F. Zhang, K. Deisseroth, and A. V. Nurmikko, “Integrated device for optical
stimulation and spatiotemporal electrical recording of neural activity in light-sensitized
brain tissue,” Journal of Neural Engineering, vol. 6, no. 5, 2009.

[87]

J. Wang, F. Wagner, D. A. Borton, J. Zhang, I. Ozden, R. D. Burwell, A. V.
Nurmikko, R. van Wagenen, I. Diester, and K. Deisseroth, “Integrated device for
combined optical neuromodulation and electrical recording for chronic in vivo
applications,” Journal of Neural Engineering, vol. 9, no. 1, p. 016001, 2011.

[88]

F. Wu, E. Stark, M. Im, I.-J. Cho, E.-S. Yoon, G. Buzsáki, K. D. Wise, and E. Yoon,
“An implantable neural probe with monolithically integrated dielectric waveguide and
recording electrodes for optogenetics applications,” Journal of Neural Engineering,
vol. 10, no. 5, p. 056012, 2013.

[89]

B. Rubehn, S. B. E. Wolff, P. Tovote, A. Lüthi, and T. Stieglitz, “A polymer-based
neural microimplant for optogenetic applications: design and first in vivo study,” Lab
on a Chip, vol. 13, no. 4, p. 579, 2013.

[90]

K. Tamura, Y. Ohashi, T. Tsubota, D. Takeuchi, T. Hirabayashi, M. Yaguchi, M.
Matsuyama, T. Sekine, and Y. Miyashita, “A glass-coated tungsten microelectrode
enclosing optical fibers for optogenetic exploration in primate deep brain structures,”
Journal of Neuroscience Methods, vol. 211, no. 1, pp. 49–57, 2012.

[91]

A. Blau, “Prospects for neuroprosthetics: flexible microelectrode arrays with polymer
conductors,” Applied biomedical Engineering, no. 5, pp. 83-122, 2011.

[92]

I. R. Minev, D. J. Chew, E. Delivopoulos, J. W. Fawcett, and S. P. Lacour, “High
sensitivity recording of afferent nerve activity using ultra-compliant microchannel
electrodes: an acute in vivo validation,” Journal of Neural Engineering, vol. 9, no. 2, p.
026005, 2012.

[93]

D. Rodger, A. Fong, W. Li, and H. Ameri, “Flexible Parylene-based multielectrode
array technology for high-density neural stimulation and recording,” Sensors and

137

Actuators B, vol. 132, pp.449-460, 2008.
[94]

W. Li, D. C. Rodger, E. Meng, J. D. Weiland, M. S. Humayun, and Y.-C. Tai, “WaferLevel Parylene Packaging With Integrated RF Electronics for Wireless Retinal
Prostheses,” Journal of Microelectromechanical Systems, vol. 19, no. 4, pp. 735–742,
2010.

[95]

W. Li, D. C. Rodger, A. Pinto, E. Meng, J. D. Weiland, M. S. Humayun, and Y.-C.
Tai, “Parylene-based integrated wireless single-channel neurostimulator,” Sensors and
Actuators A: Physical, vol. 166, no. 2, pp. 193–200, 2011.

[96]

S. Takeuchi, D. Ziegler, Y. Yoshida, K. Mabuchi, and T. Suzuki, “Parylene flexible
neural probes integrated with microfluidic channels,” Lab on a Chip, vol. 5, no. 5, p.
519, 2005.

[97]

"Parylene conformal coating specifications & properties," [Online]. Available:
http://scscoatings.com.

[98]

D. Kim, J. Viventi, J. Amsden, J. Xiao, L. Vigeland, Y. Kim, J. Blanco, B. Panilaitis,
E. Frechette, and D. Contreras, “Dissolvable films of silk fibroin for ultrathin
conformal bio-integrated electronics,” Nature Materials, vol. 9, no. 6, pp. 511–517,
2010.

[99]

J. G. McCall, T.-I. Kim, G. Shin, X. Huang, Y. H. Jung, R. Al-Hasani, F. G. Omenetto,
M. R. Bruchas, and J. A. Rogers, “Fabrication and application of flexible, multimodal
light-emitting devices for wireless optogenetics,” Nature Protocol, vol. 8, no. 12, pp.
2413–2428, 2013.

[100]

R. Pashaie, P. Anikeeva, J. H. Lee, R. Prakash, O. Yizhar, M. Prigge, D. Chander, T. J.
Richner, and J. Williams, “Optogenetic Brain Interfaces,” IEEE Reviews in Biomedical
Engineering, vol. 7, pp. 3–30, 2014.

[101]

K. Zheng, D. C. MacDonald, and K. Deisseroth, “Optogenetic and potassium channel
gene therapy in a rodent model of focal neocortical epilepsy,” Science translational
medicine, vol. 4, no. 161, pp.1-10, 2012.

[102]

V. Gradinaru, M. Mogri, K. R. Thompson, J. M. Henderson, and K. Deisseroth,
“Optical Deconstruction of Parkinsonian Neural Circuitry,” Science, vol. 324, no.
5925, pp. 354–359, Apr. 2009.

[103]

L. Petreanu, T. Mao, S. M. Sternson, and K. Svoboda, “The subcellular organization of
neocortical excitatory connections,” Nature, vol. 457, pp. 1142–1145, 2009.

[104]

R. Hira, N. Honkura, J. Noguchi, and E. al, “Transcranial optogenetic stimulation for
functional mapping of the motor cortex,” Journal of Neuroscience Methods, vol. 179,
pp. 258–263, 2009.

[105]

S. J. Cruikshank, H. Urabe, A. V. Nurmikko, and B. W. Connors, “Pathway-Specific

138

Feedforward Circuits between Thalamus and Neocortex Revealed by Selective Optical
Stimulation of Axons,” Neuron, vol. 65, no. 2, pp. 230–245, Jan. 2010.
[106]

C. P. Goold and R. A. Nicoll, “Single-cell optogenetic excitation drives homeostatic
synaptic depression,” Neuron, vol. 68, pp. 512–528, 2010.

[107]

O. Yizhar, L. E. Fenno, M. Prigge, F. Schneider, T. J. Davidson, D. J. O'Shea, V. S.
Sohal, I. Goshen, J. Finkelstein, J. T. Paz, K. Stehfest, R. Fudim, C. Ramakrishnan, J.
R. Huguenard, P. Hegemann, and K. Deisseroth, “Neocortical excitation/inhibition
balance in information processing and social dysfunction,” Nature, vol. 477, no. 7363,
pp. 171–178, 2011.

[108]

J. A. Cardin, M. Carlen, K. Meletis, U. Knoblich, F. Zhang, K. Deisseroth, L.-H. Tsai,
and C. I. Moore, “Driving fast-spiking cells induces gamma rhythm and controls
sensory responses,” Nature, vol. 459, no. 7247, pp. 663–667, Apr. 2009.

[109]

V. S. Sohal, F. Zhang, O. Yizhar, and K. Deisseroth, “Parvalbumin neurons and
gamma rhythms enhance cortical circuit performance,” Nature, vol. 459, no. 7247, pp.
698–702, 2009.

[110]

M. R. Warden, A. Selimbeyoglu, J. J. Mirzabekov, M. Lo, K. R. Thompson, S.-Y.
Kim, A. Adhikari, K. M. Tye, L. M. Frank, and K. Deisseroth, “A prefrontal cortex–
brainstem neuronal projection that controls response to behavioural challenge,”
Nature, vol. 492, no. 7429, pp. 428–432, 2012.

[111]

M. E. Carter, J. Brill, P. Bonnavion, J. R. Huguenard, R. Huerta, and L. de Lecea,
“Mechanism for Hypocretin-mediated sleep-to-wake transitions,” Proceedings of the
National Academy of Sciences of the United States of America, vol. 109, no. 39, pp.
E2635–E2644, 2012.

[112]

M. Desai, I. Kahn, U. Knoblich, J. Bernstein, H. Atallah, A. Yang, N. Kopell, R. L.
Buckner, A. M. Graybiel, C. I. Moore, and E. S. Boyden, “Mapping brain networks in
awake mice using combined optical neural control and fMRI,” Journal of
Neurophysiology, vol. 105, no. 3, pp. 1393–1405, 2011.

[113]

J. H. Lee, R. Durand, V. Gradinaru, F. Zhang, I. Goshen, D.-S. Kim, L. E. Fenno, C.
Ramakrishnan, and K. Deisseroth, “Global and local fMRI signals driven by neurons
defined optogenetically by type and wiring,” Nature, vol. 465, no. 7299, pp. 788–792,
2010.

[114]

J. H. Lee, “Tracing Activity Across the Whole Brain Neural Network with Optogenetic
Functional Magnetic Resonance Imaging,” Frontiers in Neuroinformatics, vol. 5,
2011.

[115]

I. Kahn, M. Desai, U. Knoblich, J. Bernstein, M. Henninger, A. M. Graybiel, E. S.
Boyden, R. L. Buckner, and C. I. Moore, “Characterization of the Functional MRI
Response Temporal Linearity via Optical Control of Neocortical Pyramidal Neurons,”
Journal of Neuroscience, vol. 31, no. 42, pp. 15086–15091, 2011.
139

[116]

G. B. Kim, J. R. Cho, H.-S. Shin, and J. H. Choi, “Cortical mapping of the optically
evoked responses in channelrhodopsin-2 mouse model,” Engineering in Medicine and
Biology Society (EMBC) 2011 Annual International Conference of the IEEE, pp.
6769–6772, 2011.

[117]

P. Ledochowitsch, E. Olivero, T. Blanche, and M. M. Maharbiz, “A transparent
μECoG array for simultaneous recording and optogenetic stimulation,” Engineering in
Medicine and Biology Society (EMBC) 2011 Annual International Conference of the
IEEE, pp. 2937–2940, 2011.

[118]

T. J. Richner, S. Thongpang, S. K. Brodnick, A. A. Schendel, R. W. Falk, L. A.
Krugner-Higby, R. Pashaie, and J. C. Williams, “Optogenetic microelectrocorticography for modulating and localizing cerebral cortex activity,” Journal of
Neural Engineering, vol. 11, no. 1, p. 016010, 2014.

[119]

T. Ban, Y. Ohya, and Y. Takahashi, “Tin doped indium oxide thin films: Electrical
properties,” Journal of Applied Physics, 1998.

[120]

M. Rottmann and K. H. Heckner, “Electrical and structural properties of indium tin
oxide films deposited by reactive DC sputtering,” Journal of Physics D: Applied
Physics, vol. 28, no. 7, pp. 1448–1453, 1999.

[121]

J. Selvakumaran, M. Hughes, J. Keddie, and D. Ewins, “Assessing biocompatibility of
materials for implantable microelectrodes using cytotoxicity and protein adsorption
studies,” Microtechnologies in Medicine & Biology 2nd Annual International IEEEEMB Special Topic Conference on, pp. 261–264, 2002.

[122]

M. W. Slutzky, L. R. Jordan, T. Krieg, M. Chen, D. J. Mogul, and L. E. Miller,
“Optimal spacing of surface electrode arrays for brain–machine interface applications,”
Journal of Neural Engineering, vol. 7, no. 2, p. 026004, 2010.

[123]

H. Onoe, A. Nakai, E. Iwase, K. Matsumoto, and I. Shimoyama, “Temperaturecontrolled transfer and self-wiring for multi-color light-emitting diode arrays,” Journal
of Micromechanics and Microengineering, vol. 19, no. 7, 2009.

[124]

Y. S. Jeong, B. Ratier, A. Moliton, and L. Guyard, “UV–visible and infrared
characterization of poly ( p-xylylene) films for waveguide applications and OLED
encapsulation,” Synthetic metals, vol. 127, no. 1, pp. 189–193, 2002.

[125]

F. Kurdesau, G. Khripunov, A. F. da Cunha, M. Kaelin, and A. N. Tiwari,
“Comparative study of ITO layers deposited by DC and RF magnetron sputtering at
room temperature,” Journal of Non-Crystalline Solids, vol. 352, no. 9, pp. 1466–1470,
2006.

[126]

P. Andersen and E. I. Moser, “Brain temperature and hippocampal function,”
Hippocampus, vol. 5, no. 6, pp. 491–498, 1995.

[127]

K. Kwon, S. Eldawlatly, and K. Oweiss, “NeuroQuest: A comprehensive analysis tool

140

for extracellular neural ensemble recordings,” Journal of Neuroscience Methods, vol.
204, no. 1, pp. 189–201, 2012.
[128]

A. Zorzos, A. Dietrich, G. Franzesi, B. Chow, X. Han, C. Fonstad, and E. Boyden,
“Lightproof neural recording electrodes,” Medicine and Biology Society (EMBC) 2009
Annual International Conference of the IEEE, 2009.

[129]

H. Huang and E. al, “Different fabrication methods of out-of-plane polymer hollow
needle arrays and their variations,” Journal of Micromechanics and Microengineering,
vol. 17, no. 2, pp. 393–402, 2007.

[130]

K. Kwon, X. Bi, and W. Li, “Droplet Backside Exposure for Making Slanted SU-8
Microneedles,” Engineering in Medicine and Biology Society (EMBC) 2013 Annual
International Conference of the IEEE, 2013.

[131]

Optical Fiber Commn. McGraw-Hill Education (India) Pvt Limited, 2008.

[132]

W.-C. Yang, Y.-C. Chen, Y.-S. Huang, Y.-Y. Fu, S.-C. Tang, and C.-C. Fu,
“Engineering a Biomimetic Villus Array for In Vitro Three-Dimensional Culture of
Intestinal Epithelial Cells,” Journal of Microelectromechanical Systems., vol. 21, no.
6, pp. 1418–1425, 2012.

[133]

J.-H. Park, Y.-K. Yoon, S.-O. Choi, M. R. Prausnitz, and M. G. Allen, “Tapered
Conical Polymer Microneedles Fabricated Using an Integrated Lens Technique for
Transdermal Drug Delivery,” IEEE Transactions on Biomedical engineering, vol. 54,
no. 5, pp. 903–913, 2007.

[134]

S. P. Davis, B. J. Landis, Z. H. Adams, M. G. Allen, and M. R. Prausnitz, “Insertion of
microneedles into skin: measurement and prediction of insertion force and needle
fracture force,” Journal of Biomechanics, vol. 37, no. 8, pp. 1155–1163, 2004.

[135]

D. V. McAllister, M. G. Allen, and M. R. Prausnitz, “Microfabricated microneedles for
gene and drug delivery,” Annual Review of Biomedical Engineering, vol. 2, no. 1, pp.
289–313, 2000.

[136]

D. V. McAllister, P. M. Wang, S. P. Davis, J. H. Park, P. J. Canatella, M. G. Allen, and
M. R. Prausnitz, “Microfabricated needles for transdermal delivery of macromolecules
and nanoparticles: fabrication methods and transport studies,” Proceedings of the
National Academy of Sciences of the United States of America, vol. 100, no. 24, pp.
13755–13760, 2003.

[137]

A. Branner and R. A. Normann, “A multielectrode array for intrafascicular recording
and stimulation in sciatic nerve of cats,” Brain Research Bulletin, vol. 51, no. 4, pp.
293–306, 2000.

[138]

J.-H. Park, M. G. Allen, and M. R. Prausnitz, “Biodegradable polymer microneedles:
Fabrication, mechanics and transdermal drug delivery,” Journal of Controlled Release,
vol. 104, no. 1, pp. 51–66, 2005.

141

[139]

H. J. G. E. Gardeniers, R. Luttge, E. J. W. Berenschot, M. J. de Boer, S. Y. Yeshurun,
M. Hefetz, R. van-t Oever, and A. van den Berg, “Silicon micromachined hollow
microneedles for transdermal liquid transport,” Journal of Microelectromechanical
Systems, vol. 12, no. 6, pp. 855–862, 2003.

[140]

S. Henry, D. V. McAllister, M. G. Allen, and M. R. Prausnitz, “Microfabricated
microneedles: A novel approach to transdermal drug delivery,” Journal of
pharmaceutical sciences, vol. 88, no. 9, p. 948, 1999.

[141]

K. H. Kim, D. Park, H. Lu, W. Che, and E. al, “A tapered hollow metallic microneedle
array using backside exposure of SU-8,” Journal of Micromechanics and
Microengineering, vol. 14, pp. 597-603, 2004.

[142]

A. Branner, R. B. Stein, and R. A. Norman, “Selective Stimulation of Cat Sciatic
Nerve Using an Array of Varying-Length Microelectrodes,” Journal of
Neurophysiology, vol. 85, pp. 1585–1594, 2013.

[143]

W.-C. Yang, Y.-S. Huang, B.-Y. Shew, and C.-C. Fu, “Study on diffraction effect and
microstructure profile fabricated by one-step backside lithography,” Journal of
Micromechanics and Microengineering, vol. 23, no. 3, p. 035004, 2013.

[144]

T. A. Duncombe, J. F. Parsons, and K. F. Bo x0308 hringer, “Droplet transport on flat
chemically heterogeneous surfaces via periodic wetting barriers and vibration,” Micro
Electro Mechanical Systems (MEMS), 2010 IEEE 23th International Conference on,
pp. 1043–1046, 2010.

[145]

S. Vafaei and M. Z. Podowski, “Theoretical analysis on the effect of liquid droplet
geometry on contact angle,” Nuclear Engineering and Design, vol. 235, no. 10, pp.
1293–1301, 2005.

[146]

V. Fakhfouri, N. Cantale, G. Mermoud, J. Y. Kim, D. Boiko, E. Charbon, A. Martinoli,
and J. Brugger, “Inkjet printing of SU-8 for polymer-based MEMS a case study for
microlenses,” Micro Electro Mechanical Systems (MEMS), 2008 IEEE 21th
International Conference on, pp. 407-410, 2008.

[147]

H. Lorenz, M. Despont, N. Fahrni, N. LaBianca, P. Renaud, and P. Vettiger, “SU-8: a
low-cost negative resist for MEMS,” Journal of Micromechanics and
Microengineering, vol. 7, no. 3, pp. 121–124, 1999.

[148]

“SU-8 3000 Data Sheet ver 4.2,” [Online]. Available: http://www.microchem.com.

[149]

G. J. Suaning, N. H. Lovell, K. Schindhelm, and M. T. Coroneo, “The bionic eye
(electronic visual prosthesis): A review,” Australian and New Zealand Journal of
Ophthalmology, vol. 26, no. 3, pp. 195–202, 1998.

[150]

"Prevalence of Adult Vision Impairment and Age-Related Eye Diseases in America,"
National Eye Institute, (Based on 2010 U.S. Census populations).

142

[151]

"The Economic Impact of Vision Problems," Prevent Blindness American
(http://www.preventblindness.net/site/DocServer/Impact_of_Vision_Problems.pdf),
Accessed on July 27, 2014.

[152]

P. Walter, G. Roessler, T. Laube, C. Brockmann, T. Kirschkamp, B. Mazinani, M.
Goertz, C. Koch, I. Krisch, B. Sellhaus, H. K. Trieu, J. Weis, N. Bornfeld, H. Rothgen,
A. Messner, and W. Mokwa, "Implantation and Explantation of a Wireless Epiretinal
Retina Implant Device: Observations during the EPIRET3 Prospective Clinical Trial,"
Investigative Ophthalmology & Visual Science, vol. 50, pp. 3003-3008, 2009.

[153]

S. Klauke, M. Goertz, S. Rein, D. Hoehl, U. Thomas, R. Eckhorn, F. Bremmer, and T.
Wachtler, "Stimulation with a Wireless Intraocular Epiretinal Implant Elicits Visual
Percepts in Blind Humans," Investigative Ophthalmology & Visual Science, vol. 52,
pp. 449-455, 2011.

[154]

Y. T. Wong, N. Dommel, P. Preston, L. E. Hallum, T. Lehmann, N. H. Lovell, and G.
J. Suaning, “Retinal Neurostimulator for a Multifocal Vision Prosthesis,” Neural
Systems and Rehabilitation Engineering, IEEE Transactions on, vol. 15, no. 3, pp.
425–434, 2007.

[155]

M. S. Humayun, J. D. Dorn, A. K. Ahuja, A. Caspi, E. Filley, G. Dagnelie, J.
Salzmann, A. Santos, J. Duncan, L. daCruz, S. Mohand-Said, D. Eliott, M. J.
McMahon, and R. J. Greenberg, "Preliminary 6 month results from the argustm ii
epiretinal prosthesis feasibility study," in Engineering in Medicine and Biology Society
(EMBC) 2009 Annual International Conference of the IEEE, pp. 4566-4568, 2009.

[156]

T. Stieglitz, W. Haberer, C. Lau, and M. Goertz, "Development of an inductively
coupled epiretinal vision prosthesis," in Engineering in Medicine and Biology Society
(EMBC) 2004. IEMBS '04. 26th Annual International Conference of the IEEE, pp.
4178-4181, 2004.

[157]

M. S. Humayun, J. D. Dorn, L. da Cruz, G. Dagnelie, J. A. Sahel, P. E. Stanga, A. V.
Cideciyan, J. L. Duncan, D. Eliott, E. Filley, A. C. Ho, A. Santos, A. B. Safran, A.
Arditi, L. V. Del Priore, and R. J. Greenberg, "Interim Results from the International
Trial of Second Sight's Visual Prosthesis," Ophthalmology, vol. 119, pp. 779-788,
2012.

[158]

A. Y. Chow, V. Y. Chow, K. H. Packo, J. S. Pollack, G. A. Peyman, and R. Schuchard,
"The artificial silicon retina microchip for the treatment of vision loss from retinitis
pigmentosa," Archives of Ophthalmology, vol. 122, pp. 460-469, 2004.

[159]

E. Zrenner, K. U. Bartz-Schmidt, H. Benav, D. Besch, A. Bruckmann, V.-P. Gabel, F.
Gekeler, U. Greppmaier, A. Harscher, S. Kibbel, J. Koch, A. Kusnyerik, T. Peters, K.
Stingl, H. Sachs, A. Stett, P. Szurman, B. Wilhelm, and R. Wilke, "Subretinal
electronic chips allow blind patients to read letters and combine them to words,"
Proceedings of the Royal Society B: Biological Sciences, vol. 278, pp. 1489-1497,
2011.

143

[160]

J. F. Rizzo, J. Wyatt, J. Loewenstein, S. Kelly, and D. Shire, "Perceptual Efficacy of
Electrical Stimulation of Human Retina with a Microelectrode Array during ShortTerm Surgical Trials," Investigative Ophthalmology & Visual Science, vol. 44, pp.
5362-5369, 2003.

[161]

G. S. Brindley and W. S. Lewin, "The sensations produced by electrical stimulation of
the visual cortex," The Journal of Physiology, vol. 196, pp. 479-493, 1968.

[162]

W. H. Dobelle and M. G. Mladejovsky, “Phosphenes produced by electrical
stimulation of human occipital cortex, and their application to the development of a
prosthesis for the blind,” The Journal of physiology, vol. 243, no. 2, pp. 553–576,
1974.

[163]

E. Margalit, M. Maia, J. D. Weiland, R. J. Greenberg, G. Y. Fujii, G. Torres, D. V.
Piyathaisere, T. M. O'Hearn, W. Liu, G. Lazzi, G. Dagnelie, D. A. Scribner, E. de Juan
Jr, and M. S. Humayun, "Retinal Prosthesis for the Blind," Survey of Ophthalmology,
vol. 47, pp. 335-356, 2002.

[164]

E. J. Tehovnik, W. M. Slocum, S. M. Smirnakis, and A. S. Tolias, "Microstimulation
of visual cortex to restore vision," in Progress in Brain Research, vol. 175, pp. 347375, 2009.

[165]

E. M. Schmidt, M. J. Bak, F. T. Hambrecht, C. V. Kufta, D. K. O'rourke, and P.
Vallabhanath, “Feasibility of a visual prosthesis for the blind based on intracortical
micro stimulation of the visual cortex,” Brain, vol. 119, no. 2, pp. 507–522, 1996.

[166]

R. A. Normann, B. A. Greger, P. House, S. F. Romero, F. Pelayo and E. Femandez,
“Toward the development of a cortically based visual neuroprosthesis,” Journal of
Neural Engineering, vol. 6, p. 035001, 2009.

[167]

P. Troyk, M. Bak, J. Berg, D. Bradley, S. Cogan, R. Erickson, C. Kufta, D. McCreery,
E. Schmidt, and V. Towle, "A Model for Intracortical Visual Prosthesis Research,"
Artificial Organs, vol. 27, pp. 1005-1015, 2003.

[168]

M. Jazayeri, Z. Lindbloom-Brown, and G. D. Horwitz, “Saccadic eye movements
evoked by optogenetic activation of primate V1,” Nature Neuroscience, vol. 15, no.
10, pp. 1368–1370, 2012.

[169]

M. Jazayeri, Z. Lindbloom-Brown, and G. D. Horwitz, "Phosphene induction in
monkeys using optogenetics," in Computational and Systems Neuroscience (COSYNE
2012), Salt Lake City, UT, February 23-26, 2012.

[170]

H.-M. Lee and M. Ghovanloo, “A power-efficient wireless capacitor charging system
through an inductive link,” IEEE Trans. on Circuits and Systems-II, vol. 60, no. 10, pp.
707-711, 2013.

[171]

M. Histed, V. Bonin, and R. Reid, “Direct activation of sparse, distributed populations
of cortical neurons by electrical microstimulation,” Neuron, vol. 63, no. 4, pp. 508–

144

522, 2009.
[172]

A. Stroh, H. Adelsberger, A. Groh, C. Rühlmann, S. Fischer, A. Schierloh, K.
Deisseroth, and A. Konnerth, “Making Waves: Initiation and Propagation of
Corticothalamic Ca2+ Waves In Vivo,” Neuron, vol. 77, no. 6, pp. 1136–1150, 2013.

[173]

R. Hira, F. Ohkubo, Y. R. Tanaka, Y. Masamizu, G. J. Augustine, H. Kasai, and M.
Matsuzaki, “In vivo optogenetic tracing of functional corticocortical connections
between motor forelimb areas,” Front. Neural Circuits, vol. 7, 2013.

[174]

J. Jacobs and M. J. Kahana, “Direct brain recordings fuel advances in cognitive
electrophysiology,” Trends in Cognitive Sciences, vol. 14, no. 4, pp. 162–171, 2010.

[175]

T. Womelsdorf, J. M. Schoffelen, R. Oostenveld, and W. Singer, “Modulation of
Neuronal Interactions Through Neuronal Synchronization,” Science, vol. 316, pp.
1609-1612, 2007.

[176]

N. Kopell, G. B. Ermentrout, M. A. Whittington, and R. D. Traub, “Gamma rhythms
and beta rhythms have different synchronization properties,” Proceedings of the
National Academy of Sciences of United States of America, vol. 97, no. 4, pp.19671872, 2000.

[177]

S. Aviyente and A. Y. Mutlu, “A Time-Frequency-Based Approach to Phase and Phase
Synchrony Estimation,” IEEE Transactions on Signal Processing, vol. 59, no. 7, pp.
3086–3098, 2011.

[178]

E. Pereda, R. Q. Quiroga, and J. Bhattacharya, “Nonlinear multivariate analysis of
neurophysiological signals,” Progress in Neurobiology, vol. 77, no. 1, pp. 1–37, 2005.

[179]

M. Rosenblum, A. Pikovsky, and J. Kurths, “Phase Synchronization of Chaotic
Oscillators,” Physical Review Letters, vol. 76, no. 11, pp. 1804–1807, 1996.

[180]

J. Jacobs and M. J. Kahana, “Neural representations of individual stimuli in humans
revealed by gamma-band electrocorticographic activity,” The Journal of Neuroscience,
vol. 29, no. 33, pp. 10203-10214, 2009.

[181]

C. T. Wentz, J. G. Bernstein, P. Monahan, A. Guerra, A. Rodriguez, and E. S. Boyden,
“A wirelessly powered and controlled device for optical neural control of freelybehaving animals,” Journal of Neural Engineering, vol. 8, no. 4, p. 046021, 2011.

[182]

A. J. Heller, "Classification and epidemiology of tinnitus," Otolaryngologic Clinics of
North America, vol. 36, pp. 239-248, 2003.

[183]

J. G. Turner, T. J. Brozoski, C. A. Bauer, J. L. Parrish, K. Myers, L. F. Hughes, and D.
M. Caspary, “Gap detection deficits in rats with tinnitus: A potential novel screening
tool.,” Behavioral Neuroscience, vol. 120, no. 1, pp. 188–195, 2006.

[184]

J. J. Eggermont and L. E. Roberts, “The neuroscience of tinnitus,” Trends in

145

Neurosciences, vol. 27, no. 11, pp. 676–682, 2004.
[185]

N. Ahmad and M. Seidman, “Tinnitus in the Older Adult - Springer,” Drugs & aging,
vol. 21, pp. 297-305, 2004.

[186]

R. Salvi, E. Lobarinas, and W. Sun, "Pharmacological treatments for tinnitus: new and
old," Drugs of the Future, vol. 34, pp. 381-400, 2009.

[187]

U. Rosenhall and A.-K. Karlsson, "Tinnitus in old age," Scandinavian audiology, vol.
20, pp. 165-171, 1991.

[188]

S. Griest and P. Bishop, "Tinnitus as an early indicator of permanent hearing loss. A 15
year longitudinal study of noise exposed workers," AAOHN journal: official journal of
the American Association of Occupational Health Nurses, vol. 46, pp. 325-329, 1998.

[189]

A. R. Møller, “Tinnitus and pain,” Progress in Brain Research, vol. 166, pp. 47-53,
2007.

[190]

E. Lobarinas, G. Yang, W. Sun, D. Ding, N. Mirza, W. Dalby-Brown, E. Hilczmayer,
S. Fitzgerald, L. Zhang, and R. Salvi, "Salicylate-and quinine-induced tinnitus and
effects of memantine," Acta oto-laryngologica, vol. 126, pp. 13-19, 2006.

[191]

J. J. Eggermont and L. E. Roberts, "The neuroscience of tinnitus: understanding
abnormal and normal auditory perception," Frontiers in systems neuroscience, vol. 6,
pp. 1-4, 2012.

[192]

J. W. House and D. E. Brackman, “Tinnitus: surgical treatment,” Ciba Foundation
Symposium, pp.203-212, 1981.

[193]

P. J. Jastreboff and C. T. Sasaki, “An animal model of tinnitus: a decade of
development,” Otology & Neurotology, vol. 15, pp. 19-27, 1994.

[194]

R. R. Llinás, U. Ribary, D. Jeanmonod, E. Kronberg, and P. P. Mitra,
“Thalamocortical dysrhythmia: A neurological and neuropsychiatric syndrome
characterized by magnetoencephalography,” Proceedings of the National Academy of
Sciences of the United States of America, vol. 96, no. 26, pp. 15222–15227, 1999.

[195]

J. Melcher, I. Sigalovsky, J. Guinan, and R. Levine, "Lateralized tinnitus studied with
functional magnetic resonance imaging: abnormal inferior colliculus activation,"
Journal of Neurophysiology, vol. 83, pp. 1058-1072, 2000.

[196]

T. J. Brozoski, C. A. Bauer, and D. M. Caspary, “Elevated Fusiform Cell Activity in
the Dorsal Cochlear Nucleus of Chinchillas with Psychophysical Evidence of
Tinnitus,” The Journal of Neuroscience, vol. 22, no. 6, pp. 2383–2390, 2002.

[197]

E. Wallhäusser-Franke, B. Cuautle-Heck, G. Wenz, G. Langner, and C. Mahlke,
“Scopolamine attenuates tinnitus-related plasticity in the auditory cortex,”
NeuroReport, vol. 17, no. 14, pp. 1487–1491, 2006.

146

[198]

J. A. Kaltenbach, “Tinnitus: Models and mechanisms,” Hearing Research, vol. 276,
no. 1, pp. 52–60, 2011.

[199]

R. Schaette, “Tinnitus in Men, Mice (as well as other Rodents), and Machines,”
Hearing Research, 2013.

[200]

D. Jin, K. K. Ohlemiller, D. Lei, E. Dong, and L. Role, “Age-related neuronal loss in
the cochlea is not delayed by synaptic modulation,” Neurobiology of Aging, vol. 32,
pp. 2321.e13-2321.e23, 2011.

[201]

J. Bao, D. Wolpowitz, L. W. Role, and D. A. Talmage, "Back signaling by the Nrg-1
intracellular domain," The Journal of cell biology, vol. 161, pp. 1133-1141, 2003.

[202]

J. Bao, H. Lin, Y. Ouyang, D. Lei, A. Osman, T.-W. Kim, L. Mei, P. Dai, K. K.
Ohlemiller, and R. T. Ambron, "Activity-dependent transcription regulation of PSD-95
by neuregulin-1 and Eos," Nature Neuroscience, vol. 7, pp. 1250-1258, 2004.

[203]

M. Ralli, E. Lobarinas, A. R. Fetoni, D. Stolzberg, G. Paludetti, and R. Salvi,
"Comparison of salicylate and quinine induced tinnitus in rats; development, time
course and evaluation of audiological correlates," Otology & neurotology: official
publication of the American Otological Society, American Neurotology Society [and]
European Academy of Otology and Neurotology, vol. 31, p. 823, 2010.

[204]

G. Yang, E. Lobarinas, L. Zhang, J. Turner, D. Stolzberg, R. Salvi, and W. Sun,
"Salicylate induced tinnitus: behavioral measures and neural activity in auditory cortex
of awake rats," Hearing Research, vol. 226, pp. 244-253, 2007.

[205]

K. S. Kraus, S. Mitra, Z. Jimenez, S. Hinduja, D. Ding, H. Jiang, L. Gray, E.
Lobarinas, W. Sun, and R. J. Salvi, "Noise trauma impairs neurogenesis in the rat
hippocampus," Neuroscience, vol. 167, pp. 1216-1226, 2010.

147