‘ 7 >Vliautm‘1.

‘ ; .
. ._ » .

:, Sunfirammmwn .
.l . 2H9

, -1mls
. r

 

 

. “.2:
2.5.3.}:
utt‘to
5.5;?!

 

 

 

 

THESA

Abel

 

LIBRAIW
Michigan State
Unlverslty

w ' "J" “.m~-.~ J

 

 

This is to certify that the
thesis entitled

Functional MRI of Cortical Activation During

Muscle Contraction and Fatigue

presented by

Robert W. Reid

has been accepted towards fulfillment
of the requirements for

Master's Physiology

degree in

 

 

W

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6101 cJCIRC/DateDue.p65—p.15

 

 

FUNCTIONAL MRI OF CORTICAL ACTIVATION DURING MUSCLE
CONTRACTION AND FATIGUE

BY

Robert W. Reid

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements

For the degree of

MASTER OF SCIENCE

Department of Physiology

2001

ABSTRACT

FUNCTIONAL MRI OF CORTICAL ACTIVATION DURING MUSCLE

CONTRACTION AND FATIGUE

BY

Robert W. Reid

The first part of this thesis examined the effect of the magnitude of muscle force
on the intensity of functional Magnetic Resonance Imaging (HVIRI) response in the
primary motor cortex. An flVIRI study was conducted with echo planar image acquisition
using 13 volunteers who squeezed a custom built handgn'p exerciser. Subjects performed
isometric muscle contractions between periods of rest using 25%, 50% or MAX MV C.
Regions of interest (ROI) were selected in the primary motor cortex and the mean signal
intensity of the flVIRI response was determined. The second part of the thesis
investigated the effect of muscle fatigue on the flVIRI response. Subjects were asked to
perform 30-second bouts of MAX MVC, flanked by 30-second rest periods during eight
and a halfminutes of echo planar imaging.

There were two major findings of this study. First, the results show that there is a
significant effect of isometric force on the flVIRI response in the contralateral motor
cortex. Second, the results provide no evidence that the fMRI response in the motor
cortex is altered by muscle fatigue. However, further studies will be needed to

demonstrate the generality of these results to other exercise and fatigue conditions.

TABLE OF CONTENTS

List of tables ........................................................................................................... iv
List of figures ........................................................................................................... v
Chapter 1: Background and Significance. ............................................................... 1
Introduction .............................................................................................................. 1
Overview of the control of muscle activity and fatigue. ......................................... 3
The basis of functional MRI. ................................................................................... 7
Limitations of functional MRI ................................................................................. 9
Other cortical activation measurement techniques ................................................ 11
Functional MRI and motor activation .................................................................... 12
Chapter 2: Methods ................................................................................................ 15
Human subjects ...................................................................................................... 15
Force measurements .............................................................................................. 15
Exercise protocols .................................................................................................. 18
MRI acquisition. .................................................................................................... 20
Data processing ...................................................................................................... 21
Statistical considerations ....................................................................................... 28
Chapter 3: Results .................................................................................................. 29
Force Levels ........................................................................................................... 29
Region of fMRI activation ..................................................................................... 32
FMRI response to each of the exercise conditions ................................................ 36
Effects of fatigue on signal intensity ..................................................................... 44
Chapter 4: Discussion and Future Directions ........................................................ 46
References .............................................................................................................. 52

iii

LIST of TABLE S

Table 1: Summary of raw force data ..................................................................... 31
Table 2: Summary of raw voxel intensity data ...................................................... 39
Table 3: Summary of motion corrected voxel intensity data ................................ 40

iv

LIST of FIGURES

Figure 1: Summary of raw force data ...................................................................... 6
Figure 2: Calibration curve of Weight (lbs) versus Voltage (v) ............................ 16
Figure 3: Sample WINDAQ display screen of muscle force output ..................... 17
Figure 4: Sample timeline of work/rest cycles ...................................................... 19
Figure 5: Anatomical 3DSPGR MRI image of the human brain ........................... 22
Figure 6: The time course of nine neighboring voxels in the axial plane .............. 24
Figure 7: Region of Interest (ROI) of possible motor and sensory activation ....... 25
Figure 8: Seven sample reference waves showing increased periods of activation
and decreased rest periods ..................................................................................... 26
Figure 9: Example of reference wave fitted to a R01 is signal intensity time
course ..................................................................................................................... 27
Figure 10: Raw force data recorded during the 25%, 50% and maximal MVC
trials ....................................................................................................................... 30
Figure 11: Loss of force from the first two cycles to the last two cycles .............. 33

Figure 12: Signal Intensity of voxels from the primary motor area versus time for
3 different force levels ........................................................................................... 34

Figure 13: Mean intensity of voxels from the primary motor area versus time for
100 % MVC ........................................................................................................... 35

Figure 14: Motion corrected voxel intensity from the primary motor area for 3
exercise trials ......................................................................................................... 37

Figure 15: Motion corrected signal intensity of voxels from the primary motor

area during the maximal MVC trial ....................................................................... 38
Figure 16: Bar graph showing percent change in voxel intensity in raw and

motion corrected data ............................................................................................ 42
Figure 17: Simple linear regressions of individual subjects .................................. 43

Figure 18: Comparison of first 2 cycles to the last 2 cycles of the raw and motion
corrected voxel intensity data ................................................................................ 45

Chapter 1: Background and Significance.

Introduction

Functional MRI (MRI) is a magnetic resonance imaging technique that exploits
changes in tissue lH—NMR properties in order to measure a particular physiological
response. By detecting changes in signal intensity of images of the brain, this modality
can be used to monitor local changes in brain activity in real time (for a review, see
Ogawa 1998). A number of studies have characterized the ability of MRI to map the
location of activity in the primary somatosensory and motor cortex during a variety of
motor tasks. The overall aim of this thesis was to explore the use of MRI for
quantitative studies of the role of the central nervous system in the activation and
maintenance of motor activity during submaximal and fatiguing isometric exercise.

The first aim of this thesis was to examine if there is a consistent relationship
between the intensity of the fMRI response in the motor cortex versus muscle isometric
force. It is well known that increased force of voluntary muscle contractions depends on
recruitment of additional motor units, which in turn depends on increased synaptic input
from cortical areas in the brain. Surprisingly, no published MRI studies have
unambiguously shown force-dependent changes in the intensity of cortical activity during
muscle contraction. Two previous groups examined the effect of variations in the force of
handgrip exercise on the fMRI response in human subjects. The first group, who used a
dynamic weight-lifting exercise, found no effect of increased weight on the cortical
response (Ludman, Cooper et al. 1996). The second group, who used an isometric
exercise, reported increased area of fMRI activation, but only a small, statistically

insignificant change in intensity of the fMRI response (Thickbroom, Phillips et al. 1998).

Resolution of this question is a necessary prerequisite for application of MRI to the
study of fatigue.

The second aim of this thesis was to examine if the fMRI response in the motor
cortex is altered by muscle fatigue. It is generally accepted that two components can play
a role in the fatigue of muscle during exercise (Bigland-Ritchie and Woods 1984):
peripheral and central. Peripheral fatigue refers to decreased force generation in the
muscle fibers themselves, and is sometimes referred to as "muscle contractile failure"
(Porter 1981). Although not completely understood (see Fitts 1994 for a comprehensive
review of the cellular mechanisms of peripheral fatigue), this component is thought to
result from local accumulation of metabolites such as inorganic phosphate, drops in pH,
and/or depletion of ATP supply. In contrast, central fatigue is a decline in force due to
the central nervous system failing to maintain muscle activation or "central drive"
(Bigland-Ritchie and Woods 1984). Whether central or peripheral components are
dominant likely depends on the exact nature of the exercise. For example, during
relatively prolonged exercises decreased central drive may be the dominant factor
(Bellemare 1984, Bigland-Ritchie and Woods 1984, see below). In that case, assuming
that the fMRI response in the motor cortex is related to the intensity of cortical activation,
one might expect decreased fMRI response during fatigue. On the other hand, during
very intense, repetitive voluntary exercise, when peripheral factors are likely to be
dominant, central drive might increase, as the subject attempts to maintain the target
force. In that case, one might expect the cortical fMRI response to increase during

fatigue. Of course, because there is at present no widely accepted measure of central

drive, it is difficult to predict a priori the likely result for any particular exercise protocol.
Therefore, the results of this study are only a first step in addressing these issues.
Overview of the control of muscle activity and fatigue.

Before any intended muscle movement can occur, neural commands from higher
centers must translate neural information into the intended actions. Intent to perform a
motor task begins in the higher centers of the brain with the initial activity taking place in
frontal lobe regions of the cortex (Deecke, Scheid et a1. 1969, Ghez 1991, Thickbroom,
Bymes et a1. 2000). In close association with frontal cortex activity is the activation of
the premotor area, which in turn activates anatomically specific motor areas in the
primary motor cortex corresponding to the desired motor function (Ghez 1991),
Thickbroom, Bymes et a1. 2000). However, many regions throughout the brain are
involved in the processing to ensure that the executed motor activation matches the
intended behavior. For example, inhibition via the motor circuit of the basal ganglia must
be removed to allow for stimulation of the premotor cortex (Cote 1991). The cerebellum
plays a key role, receiving input pertaining to the desired motor function from the cortex,
while concurrently receiving feedback from the spinal cord and periphery about the
evolving muscle movement (Ghez 1991). The spinocerebellum processes the incoming
information to finely tune the motor activity into a fluid motion allowing proper muscle
forces to maintain the desired behavior. From the primary motor area (Brodman’s area
4), the descending corticospinal tract provides a direct pathway for muscle activation by
travelling down the contralateral side until reaching a level of the spinal cord, where
primary motor neurons synapse onto lower motor neurons and intemeurons (Kelly 1991).

Lower motor axons then travel out of the ventral roots of the spinal column to the

muscles of interest. The lower motor axons act on muscle fibers by releasing
acetylcholine at the motor end plates, leading to excitation coupling and contraction of
the motor unit. In addition, muscle spindles and other sensory organs provide feedback to
both spinal and supraspinal areas. All of these components must be working correctly
with the overall result being a motor movement that matches the intended behavior.

Given the complexity of motor control, it is not surprising that our understanding
of central factors in muscle fatigue is limited. In a review by Enoka and Stuart (Enoka
and Stuart 1992), muscle fatigue was defined as a general concept intended to denote an
acute impairment of performance that includes both an increase in the perceived effort
necessary to exert a desired force and an eventual inability to produce this force. How
fatigue occurs depends on a number of factors including the strategy of both agonist and
antagonist muscle recruitment, the speed and force of the individual muscle contractions,
and the intensity and duration of the overall activity (Enoka and Stuart, 1992).

It is relatively easy to demonstrate that peripheral factors play the major role in
fatigue during very intense exercises. For example, major drops in muscle force output
can be induced by direct electrical stimulation of isolated muscles, where central factors
could play no role (Adams et a1. 1991, Fitts 1994). Although the exact mechanism of
fatigue under these conditions is not understood, there is a good correlation between
metabolic events (e.g., inorganic phosphate accumulation, decreased pH) and the decline
in force. Because very similar force and metabolite changes are observed in muscles of
human subjects during intense, repetitive voluntary exercise (Cady et al. 1989), it is
generally assumed that peripheral, metabolic factors are largely responsible for fatigue

under those conditions.

Nonetheless, some studies using the twitch-interpolation method suggest that
decreased central drive plays some role in fatigue even during intense voluntary exercise
(Bigland—Ritchie et a1. 1986, Mckenzie et a1. 1992). In this method (Merton 1954, Hales
and Gandevia 1988), central drive is estimated from the mechanical response to
administration of an electrical stimulus to the muscle during a maximal voluntary effort.
If the stimulus results in increased force above the voluntary force, it is assumed that
some motor units were not activated by the effort. During fatigue, the relative size of the
interpolated twitch increases, suggesting decreased central drive.

During low intensity, long duration exercise, central factors may be more
important, inasmuch as there is little evidence for peripheral metabolic fatigue under
these conditions. Bigland-Ritchie and Woods have emphasized that the time for which an
exercise can be performed is inversely dependent on the intensity of the exercise (Figure
1). They suggest that central fatigue becomes increasingly dominant as the intensity of
the exercise decreases and the duration increases. If this suggestion is correct, then the
exact choice of exercise intensity and duration could profoundly alter the results of a
study such as this one. Again however, there is little direct evidence for this suggestion,
and in any case there are few previous studies upon which to base the choice of exercise

protocol for this study.

P
E5

 

Endurance Time (min)

 

l.-__.___.........-..........-...................

O
11

 

 

Figure 1: Graph of percent of maximal voluntary contraction
(% MVC) versus duration of muscle contraction

The amount of time that the %MVC can be maintained before muscle failure occurs. For

any %MVC below 15%, muscle failure does not occur. (Bigland-Ritchie and Woods
1984)

The basis of firnctional MRI.

Active brain cells require chemical energy in the form of ATP for the energy
dependant processes of ion homeostasis, neurotransmitter synthesis, and synaptic vesicle
formation (Schwartz 1991). Furthermore, glial cells near active neurons also degrade
neurotransmitters and aid in ion homeostastis, and therefore also increase their rate of
metabolism. This increased energy demand during neural activation is directly coupled
to increased local blood flow (Gyngell, Bock et al. 1996, Raichle, Grubb et a1. 1976,
Olesen 1971, Ogawa, Menon et al. 1998, Roland 1980). The regulation of local brain
blood flow is not fully understood, but is thought to depend on factors such as increased
extracellular K+, decreased cellular P02, decreased local pH, and cellular adenosine
release (Ganong 1991). This brings about an increase in permeability in the tight junction
of endothelial cells (Rapoport 1976) and localized vasodilation. As 02 and CO2 easily
pass across the blood brain barrier, a transient increase in blood flow (Grubb, Raichle et
al. 1974) allows 02 and C02 transfer, which support the metabolic needs of the brain.
Direct evidence of regional increases in brain blood flow were demonstrated over two
decades ago by monitoring increases in regional cerebral blood flow (rCBF) via 133Xe
clearance (Roland and Larsen 1976) during periods of increased neural activity (Raichle,

Grubb et al. 1976), (Olesen 1971), (Roland, Larsen et al. 1980).

It is possible to measure local brain perfusion using MRI methods, and some
MRI studies are based on direct flow measurements. However, most brain MRI studies
are based on a related phenomenon known as the BOLD (blood-oxygenation-level-
dependent) effect (Ogawa, Lee et a1. 1990, Bandettini, Wong et al. 1992, Ogawa, Menon

et a1. 1993). This phenomenon depends on the effect of oxygen on the magnetic

properties of blood. Specifically, the magnetism of the iron atom in hemoglobin is
stronger when the hemoglobin is deoxygenated compared to when it is oxygenated.
Therefore, deoxygenated blood has a higher magnetic susceptibility, and causes greater
distortion of an applied magnetic field than does oxygenated blood. Distortions of the
magnetic field around blood vessels enhance the transverse relaxation rates (1/T2 and
l/I‘2*) of the surrounding tissue, and therefore, decrease the signal intensity in T2- and
T2*-weighted MR images. (Ogawa, Lee et a1. 1990). T2 is the time constant that
characterizes the intrinsic rate of decay of transverse phase coherence after an initial
excitation of nuclei in a static magnetic field (Farrar 1971), (Shaw 1976) (Homak 2000).
The T2. time constant is the combined effect of T2 and the additional phase dispersion
due to static field distortions (Ellennan 1994). Both T2 and T2. relaxation are
exponential processes, and changes in both can occur via changes in deoxyhemoglobin

content, resulting in changes in MRI signal intensity.

Surprisingly, it has been shown that blood flow increases more than local oxygen
consumption in active regions of the brain (Fox, Raichle et a1. 1986). Therefore, P02
increases in active regions, blood becomes more oxygenated, and local MRI signal
intensity increases in the active regions. At the field strength of clinical imagers (1.5
Tesla) the effect is small (a few percent signal change) but can be reliably measured,
especially if many scans are acquired, so statistical tests with good power can be
performed. The vast majority of MRI studies are based on measurement of these BOLD-

based changes in MRI signal intensity.

Limitations of Functional MRI

Unfortunately, there are several limitations to fMRI. First, because the underlying
NMR signal is relatively weak, the spatial resolution of MRI is relatively low. This is
particularly a problem in MRI studies, in which good time resolution is also needed. The
voxel size of fMRI images acquired at 1.5 T typically ranges from 30 to 100 m3.
Therefore, even the smallest fMRI voxel includes hundreds or thousands of neurons,
which may serve many different functions. Furthermore, some minimum but unknown
fraction of the neurons within a specific region may need to change activity in order to
induce a significant, detectable blood flow change.

A related problem is that small changes in neural activation may not be detected
because they are lost amongst the normal variations in activity and signal intensity. The
"basal" state may include a great deal of excitatory and inhibitory neuronal activity, as
evidenced by the fact that the "resting" brain has a high metabolic rate and oxygen
requirement (Brust 1991). In order to detect "activation", the change in signal must be
large enough to alter the signal above this resting brain activity. Comparing one resting
region to another is not a valid possibility because blood vessel density, tissue
composition, and basal activity might be different in different regions. Therefore,
experimental design for MRI is constrained to comparisons of resting signal intensity vs.
intensity acquired during a specified task.

A second general limitation of fMRI arises from the fact that it does not directly
measure electrical activity, but instead measures a complex metabolic result of the
activity. This has two consequences. First, because the BOLD effect depends on

hemoglobin oxygenation, the effect can be saturated. Therefore, there is no assurance that

the signal changes depend linearly on local neuron activity (Bandettini, Wong et a1.
1992). Second, there may be a time delay of up to 5 to 6 seconds between the activation
of neurons and the maximal BOLD response. Even when detection of the hemodynamic
response is rapid, the image acquisition will still be separated temporally from the actual
neural events (Ellerman 1994). An investigator needs to be cautious when examining
BOLD response changes as the temporal limitation could vary in different areas of the
brain over short time courses (Blamire, Ogawa et a1. 1992), (Ellerman 1994). Detection
of activation in one brain region before another region may not necessarily mean that
both regions were not activated at the same time. This limitation of fMRI must be
considered when looking at the rate at which a region becomes active. Furthermore, in
studies aimed at relating the intensity of the fMRI response to another parameter such as
force, the task duration must be sufficient for the hemodynamic response to fully
develop. Similarly, if the study includes "rest" periods between the tasks, these must be
of sufficient duration for the response to decay back to baseline.

A third limitation to MRI is motion artifacts. Over time, some degree of motion
inevitably occurs. Head translation in normal resting subjects has shown to be less than 1
degree of rotation over an hour (Zeffiro 1996), but even this slight deviation is sufficient
to cause image artifacts. Because the BOLD signal is only a few percent change, even a
small displacement of the image voxels across regions with different intensity can have
very large effects. If the motion occurs in synchrony with the task under study, these
motion induced-changes will mimic task-related activation. Therefore, to maximize the

accuracy of fMRI, head motion needs to be kept to an absolute minimum.

10

Finally, other imaging artifacts can complicate fMRI interpretation. For example,
the echo-planar imaging method commonly used in MRI studies is particularly sensitive
to magnetic susceptibility artifacts around bones and air-filled sinuses (Savoy 2001).
These artifacts can severely distort or even eliminate signal from adjacent brain regions.
Similarly, variations in coil tuning, magnet homogeneity, and other instrument settings
alter the absolute signal intensity, making direct comparisons between subjects difficult.
Other cortical activation measurement techniques

Detection of cortical activity changes via MRI shows good correlation with other
metabolic brain measurement techniques such as Positron Emission Tomography or PET
(Ramsey, Kirkby et a1. 1996). Both techniques are able to show specific regional
activation in the cortex during an applied stimulus. PET scans have the disadvantage of
requiring injection of a tracer whereas MRI is completely noninvasive (Martin 1991).
Another technique that is increasingly used for investigating brain motor function is
transcranial magnetic stimulation or TMS (Rothwell 1997, Ilmoniemi, Ruohonen et al.
1999, Chen 2000). TMS passes a current through a wire coil placed on the scalp. A large
changing magnetic field induces current in the underlying brain to cause stimulation and
manipulation of cortical activity (Ilmoniemi, Ruohonen et a1. 1999, Chen 2000). This
technique can be combined with anatomical MRI to determine where the stimulation
actually occurs. This technique is painless unlike the traditional electrical stimulation
(TES) as performed by Merton in the early eighties (Merton and Morton 1980).
Functional MRI can monitor changes in the cortex but it does not provide means to create
cortical stimulation like TMS can. Combined, these two techniques would make a

powerful investigative tool and attempts have been made to use the two techniques in

11

concert (Ilmoniemi, Ruohonen et a1. 1999). Unfortunately, TMS is not yet widely

available.

Functional MRI and motor activation

The use of fMRI as a tool for studying motor activity has focused primarily on
mapping of the motor cortex. Bandettini et a1. (Bandettini, Wong et a1. 1992) were among
the first to use mm to demonstrate that changes in intensity in the primary motor area
correlated with a motor task. Further studies mapped both the primary sensory (Roberts
and Rowley 1997) and motor areas (Yetkin, Mueller et a1. 1995, Boroojerdi, Foltys et a1.
1999, Kim, Ashe et al. 1993, Rao et a1. 1993). More recent studies have focused on
mapping regions involved in motor planning, including studies of fairly complex motor
tasks (Thickbroom, Bymes et a1. 2000, Koechlin, Corrado et a1. 2000). Mapping of the
motor cortex using fMRI could be of clinical use as a possible diagnostic tool before
brain surgery, and accordingly much of the current work is focusing on adaptations to
stroke and other pathologies.

Although use of fMRI to characterize the location of neural activation during a
motor task has been well documented, little attention has been paid to the intensity of the
activation. Thus far, only three papers have attempted to investigate what happens to the
intensity of the motor fMRI response during muscle contractions. The first was the study
by Ludman et a1. (Ludman, Cooper et a1. 1996) that used a dynamic finger flexion
weight-lifting exercise with heavy vs. light weights. This study found that a five-fold
increase in weight lifted had no effect on the intensity of the fMRI response in the
sensorimotor cortex. However, this study is not conclusive, because the use of a dynamic

exercise to investigate the fMRI response adds a major tactile and proprioceptive

12

component, which could cause a similar pattern in the fMRI response regardless of the
force of muscle contraction (Thickbroom et a1. 1999).

In the second paper investigating the effect of force on signal intensity,
Thickbroom et a1. (Thickbroom, Phillips et al. 1998) measured the fMRI response to a
series of isometric handgrip exercises at different force levels. They found that an
increase in the total cortical area occurred with increases in muscle force, but did not find
a significant increase in signal intensity. In this study, the highest level of force used was
50% MVC, which may not be a sufficient force to illicit the maximum change in voxel
intensity. Also, subjects were required to use four fingers to squeeze a modified
sphygmomanometer bladder attached to a mercury column. This type of exercise could
exclude activation from the thumb at lower levels of MVC, but at 50% MVC, subjects
may have used their thumb muscles to increase stabilization. If this were so, then the
observed increase in total cortical area of activation is not surprising, because the thumb
has a relatively large representation in the motor homunculus. Furthermore, as subjects
squeezed the bladder, they were required to watch the mercury column to maintain a
constant level of force. A possible downside to this type of exercise protocol is that
cognitive effort would be required to interpret and constantly correct the level of force.
This cognitive effort could possibly mask any detectable fMRI changes due to muscle
force per se. As noted below, we removed the feedback component of the exercise, and
instead had each subject practice reaching the target force prior to the fMRI imaging
session. Finally, the use of a fluid filled bladder device has another drawback. The initial
squeezing of such a device could be dynamic in nature and require finger flexion up to

the point where sufficient resistance occurs. In fact, in their more recent paper,

13

Thickbroom et al. found that the force and frequency of dynamic muscle contractions did
not correlate with the fMRI response, but the dynamic exercise had a more pronounced
effect on MRI response than isometric exercises of similar force (Thickbroom, Phillips
et a1. 1999). Therefore, it may be that isometric exercise is more appropriate for studies
of the effect of force on cortical fMRI signal intensity. For example, a previous PET
study did find force-correlated changes in blood flow during isometric exercise (Dettrners
1995).

Based on the results of these previous studies, it appears that the relationship
between the intensity of the fMRI response vs. force has not been clearly established.
Therefore, the first aim of the following experiments was to examine if variations in
isometric force alter fMRI intensity in the motor cortex. In addition, we began the

exploration of the possible effect of fatigue on the fMRI response.

14

Chapter 2: Methods

Human Subjects
Thirteen subjects (age 18 to 30 years, 5 female) were recruited from the academic

community. The study was approved by the University Committee on Research
Involving Human Subjects, and subjects were informed about the purpose and procedures
of the experimental protocol and gave written consent prior to participation. Subjects lay
prone in the magnet throughout the experiments. Due to the confining nature of the
imaging device, subject applicants who reported previous claustrophobic episodes were
excluded from the study.
Force Measurements

The force of isometric handgrip contractions was measured using a custom-built
strain gauge transducer device. This device consisted of a commercial plastic handgrip
exerciser, fitted with a 1/4‘h-inch thick aluminum bar to which two semiconductor strain
gauges where glued (Omega Engineering Inc., model 213.19-2004, 120 ohms, Stamford,
CT). The gauges were arranged in a Wheatstone bridge, and bridge excitation and signal
amplification were provided by a Gould transducer amplifier. The device was calibrated
with weights, and its output was linearly dependent on weight in the range from 2 to 16
pounds (Figure 2). The calibration was checked prior to each new study by hanging a
known weight of 2.23 Kg (10 lbs.) from the end of the hand grip and adjusting the
amplifier gain to 1 volt output. During the experiments force was digitally recorded at
120 samples/s on a MS Windows 98 PC using a 12 bit AID converter and serial data

acquisition system

15

Hand grip calibration graph

 

 

 

 

 

 

 

 

3
0 Weight (lbs) vs Volts (v)
— r2 = 0.998
2 _
E
,9
'5
>
1 l
0 l l l l l l I
O 2 4 6 8 10 12 14 16

Weight (lbs)

Figure 2: Calibration curve of Weight (lbs) versus Voltage (v)

Weights were added in random order to ensure that the handgrip device returns to starting
position and that hysteresis is not occurring. Measurements were repeated 3 times.

16

 

Figure 3: Sample WINDAQ display screen of muscle force
output (DATAQ Instruments, Akron, OH)

The green line represents the amount of actual force being produced when squeezing the
handgrip dynarnometer. The yellow line represents the target level that subjects practiced
trying to attain during each practice session prior to each imaging session. The above
figure is on color.

(DATAQ Instruments, Model DI-151RS, Akron, OH). When appropriate (see below) a
real-time display of force output was presented to the subject in the magnet on a 640x480
pixel LCD (IFIS MRI display system, MRI Devices Corporation, Waukesha, WI) using
WINDAQ Lite software (DATAQ Instruments, see Figure 3).
Exercise protocols

The force of maximum voluntary contraction (MVC) was measured for each
subject at the start of each study. MVC was measured as the mean of the peak force
during the best two of 3-6 test contractions, each 5-10 seconds in duration, and recorded
while the subject lay prone in the same position as during the MRI acquisitions. Based on
the MVC measurement, forces corresponding to 25% and 50% MVC for each subject
were computed. The subject then performed three exercise protocols, during which fMRI
(see below) and force data was continuously acquired: 25%, 50%, and MAX exercises.
The 25% and 50% exercises each consisted of four 30 s duration isometric contractions at
that force level, separated and flanked by seven 30 3 rest periods (4.5 min total duration,
see Figure 4A). Prior to the 25% and 50% exercises the subject was allowed to practice
achieving the target force by viewing the force record, along with a yellow trace
corresponding to the target force, on the IFIS display (Figure 3). However, in order to
simplify the motor task, force was not displayed to the subject during the experimental
runs. Instead, the subjects were instructed to focus their attention on minimizing head
motion (see below), and to contract only during the 30 8 "WORK" phase of an alternating
"REST"-"WORK" cue presented on the IFIS display. The 25% and 50% exercises were
performed in random order, followed by the MAX exercise. The MAX exercise consisted

of eight 30 s duration contractions, flanked by nine 30 3 rest periods (8.5 min, see Figure

18

 

 

Part A

Work Work Work Work

Rest I l I I l I I I

PartB /
Restllllllllllllllll

Time

 

Figure 4: Sample timeline of work/rest cycles during the 25%,
50% and 100% MVC protocols

Part A represents the reference waveform for 25 and 50% MVC. Part B represents the
protocol for MAX MVC, which involved 8 cycles of work/rest rather tlmn 4 cycles. The
above figure is on color.

19

 

43). Prior to this run, the subjects were instructed to contract with maximum force during
each "WORK" cue, but no practice contractions were performed. Subjects were
instructed to contract "as hard as you can", consistent with maintaining head stability.
Subjects were warned that this was likely to result in fatigue, but that they should
continue to contract as hard as possible through the 8th contraction, unless unable to
continue because of pain or discomfort. All subjects were able to complete the 8th
contraction cycle.

Prior to each exercise the subjects were cautioned to avoid head motion. In order
to aid the subjects in this task, we adopted the visual motion-sensing method described by
Thulbom et a1. (Thulbom 1999). A black string was taped horizontally across the
subject’s field of view halfway between the subject and the IFIS display screen. A
reference line was then positioned on the IFIS display so that it was aligned with the
string from the subject's perspective. Small motions of the head were then readily
apparent to the subject from the displacement of the string relative to the reference line.
MRI acquistion

MR images were acquired on a GB 1.5 Tesla clinical scanner (Horizon ver. 8.4,
GE Medical Systems, Milwaukee, WI) using a standard clinical bird-cage head coil.
After subjects were comfortably positioned, landmarked and moved into the magnet, a
Tl-weighted saggital localizer scan was acquired (spin echo, minimum TE, TR: 500 ms,
twenty 5 mm slices separated by 5 mm, 24 cm field-of-view, 256x160 acquisition matrix,
2 min total acquisition time). A high resolution anatomical 3D volume image of the
entire brain region was then acquired while the subject was at rest (spoiled gradient-

recalled echo, minimum TE and TR, 30 degree flip angle, 24 cm field of view, 1.5 mm

20

slice thickness, 124 slices, 256x128 acquisition matrix, 7 min total acquisition time).
Functional images were then continuously acquired during each of the above three
exercise protocols. The functional images were axial, one-shot, gradient-echo, echo-
planar images with TR=3 s, TE=45 ms, 90 degree flip angle, 7 mm slice thickness with
1.5 mm separation, 20 slices, 24 cm field-of-view, and 64x64 acquisition matrix. A total
of 1800 and 3400 images, or 90 and 170 images per slice, were acquired during the 25-
50% and MAX exercise protocols, respectively. After each experimental session, both
force and image data were transferred via the intemet to a Sun workstation for later
processing.
Data processing

Analysis of the fMRI data was performed in two steps: region-of-interest (ROI)
selection and motion correction, followed by fitting of an idealized waveform to the ROI
intensity variations. The ROI selection and motion correction was performed using the
AFNI software package written and distributed by Dr. Robert Cox (Cox 1996,
http://afni.nimh.nih.gov/afni). This is a general fMRI analysis package that includes
software for display of anatomical and functional data sets, various statistical functions
(e.g., cross-correlation, F-tests, multiple regression models), 2D and 3D motion
correction algorithms, Talairach transformation, and overlay of functional and anatomic
images (e. g. Figure 5). In each subject, an activated ROI was selected in the contralateral
(typically left) sensory-motor area. Activated voxels were identified by cross-correlation
against a smoothed on-off reference wave corresponding to the rest-work periods

depicted in Figure 4. The AFNI deconvolution plug-in was also used to determine if and

21

 

B

Figure 5: Anatomical 3DSPGR MRI image of the human brain

Panel A represents the axial plane. Panel B represents the coronal plane. The color
overlay represents functional regions that correlate with the reference wave. The above
figure is on color.

22

where statistical changes from resting conditions had occurred. The plug-in compares
each voxel to the reference wave and determines if a significant change from the resting
baseline has occurred during the on periods. Voxels demonstrating a high statistical
difference (p<0.001) were considered for inclusion in the ROI. An example of the signal
changes in single voxels appears in Figure 6. A contiguous subset of these voxels (mean
number = 13.7 :1: 2.3 (SE)) which were judged to be within or near the precentral motor
area was then selected as the ROI for that that subject (see figure 7). This selection was
made on the data from the first exercise protocol (i.e., randomly on the 25% or 50%
exercise data). The same ROI was then applied to the other two data sets from the same
subject. In addition, the same ROI was applied to the three data sets after motion
correction using the 3D AFNI plugin. After ROI selection, the averaged time course of
the signal intensity in the selected voxels was saved for further analysis.

To determine the relative increase in signal intensity in an unbiased manner, the
ROI intensity—time data was fit to a series of 7 basic reference waveforms (Figure 8).
These waveforms differed in the width of the "on" phase, in order to account for the fact
that in many cases the hemodynamic response persisted well beyond the 30 s contraction
duration (e.g., Figure 9). The fit was performed by linear correlation of the reference
waves vs. the actual data. The intercept of this correlation yields the fitted baseline
intensity, and the slope yields the mean change in intensity from baseline during the on
periods:

Yi = Slope * Xi + baseline,

where Yi is the intensity of the 1"“ datum point, and Xi is the intensity of the

corresponding point in the reference wave. The percent change in intensity is then

23

 

 

572. 33 . 1
[+111 3333] ’

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X: 40
Y: 36 Grid: 20 Scale: 1.5 pix/data:
z: 17 Run: 90 Baee: separate

Figure 6: The time course of nine neighboring voxels in the
axial plane

The above figure represents 9 individual graphs of 9 neighboring voxels within the
primary motor cortex during an image acquisition of alternating exercise and rest periods.
Each square represents a time series for one voxel. All neighboring voxels appear around
the highlighted center voxel. The center voxel shows a pattern similar to the reference
waves from Figure 4A, while the voxel in the upper left panel shows little similarity. The
above figure is on color.

24

 

Figure 7: Region of Interest (ROI) of possible motor and
sensory activation.

Yellow area represents the ROI. The R01 is laid over the region of activation. The above
figure is on color.

25

7 reference waves used in correlation analysis

 

 

0 20 4O 60 80 1 00

Image Number

Figure 8: Seven sample reference waves showing increased
periods of activation and decreased rest periods

The bottom most reference wave follows the same time course as the reference wave
from figure 4A. Progressing upwards, each reference wave adds one image time point to

the “WORK” cycle and removes one image data point from the “REST” cycle. A
hemodynamic delay is also added to the wave forms

26

Fitted reference wave to 50% time course

for a single subject
1180

 

 

+ Voxel intensity data
--0 -- Reference wave #4

 

 

 

 

1175 -*

1170i

 

1165 -

 

Voxel intensity

 

1160~

1155 ‘

 

 

1 1 50 I I I
0 50 1 00 1 50 200 250 300

Time (s)

 

Figure 9: Example of reference wave fitted to a ROI is signal
intensity time course

For this subject, the reference wave consisted of 7 “resting” data points and 13 “active”
data points. A shift of lhas been applied in the right direction.

27

computed as (100* slope/baseline). This correlation was computed for the seven basic
reference waves, and for the same basic waves shifted by up to 7 points in either direction
(i.e., total of 105 trials). The best fit was the variation with the best correlation
coefficient.

The MAX exercise data were analyzed in two ways. First, the first 4 rest-
contraction cycles were analyzed exactly as the 25% and 50% exercises, using the
reference waves in Figure 8. In addition, to determine if muscle fatigue affects the fMRI
response, the average intensity of first two work cycles of the maximal MVC trial were
compared to the signal intensities of the final two work cycles. This computation was
performed as above, but using truncated versions of the 7 basic reference waves.

Finally, this same analysis program was applied to the muscle force records to determine

the average resting force levels and the average force during the contractions.

Statistical considerations

One way AN OVA, paired Student's t-tests, and Leven's test of homogeneity of
variance were used to compare percent change in ROI intensity between the 3 different
conditions using SPSS statistical software (SPSS, Inc.). After ANOVA, Tukey range

tests were performed to determine which if any conditions differed from each other.

28

Chapter 3: Results

All 13 subjects succeeded in performing the 3 exercise tasks. Subjects often
reported hand soreness, which is consistent with the development of localized muscle
fatigue. Of the 13 subjects, 12 were right-handed and one left-handed. ROI analysis for
the left handed subject was performed on the contralateral motor cortex, i.e., the right
side of the brain. However, the left-handed subject along with two other subjects, were
eventually excluded from the final data analysis because of excessive head-motion. Two
additional subjects were removed from the final analysis because no detectable fMRI

response was found when examining the AFNI functional-anatomical overlay.
Force Levels

The mean MVC for the eight subjects was 14.3122 kg. Mean force recordings
(n=8) during the 25%, 50% and MAX trials are depicted in Figure 10. Contrary to the
instructions given to the subjects, force was not constant during the 30 s contractions.
Instead, a peak in force occurred at all three levels of force during the initial part of the
contractions. This phenomenon was more pronounced in the 50% and MAX exercises
than in the 25% exercise. The average decline from the peak force in the 25%, 50% and
MAX contractions was 19%, 28% and 26%, respectively. In order to obtain an unbiased
estimate of the mean force during these contractions, we applied the correlation program
described in Chapter 2 (see Table 1). Based on that analysis, the mean force during these
exercises was 19.912.0 (SE, n=8), 35.9134, and 54615.0 % MVC for the 25%, 50%,

and the first four contractions of the MAX exercise, respectively. Clearly, these subjects

29

did not on average perform as directed during the MAX exercise. Only one of the eight

subjects managed to reach an average % MVC

Force recordings at 3 different MVC levels

 

 

 

 

 

 

     

 

 

 

3

—0— MAX
—7— 25%

2

£9.

'5

>

0 1 00 200 300 400 500

Time (s)

Figure 10: Raw force data recorded during the 25%, 50% and
maximal MVC trials

Data from the 25,50 and MAX% MVC trials (N=8).

30

Table 1: Summary of raw force data

 

 

 

 

 

 

 

 

 

 

 

25% MVC DATA

Subject Shift Correlation Rest Avg. Average Reference Percent
Force (v) wave MVC
1 1 0.93 0.05 0.38 1 9.05
2 0 0.77 -0.08 0.64 1 16.00
4 1 0.88 0.18 0.91 2 28.44
6 1 0.9378 0.06 0.28 1 20.00
7 0 0.95 0.1 0.17 1 15.45
10 2 0.93 0.07 0.88 1 20.95
1 1 0 0.912 -0.03 0.25 1 25.00
12 -1 0.94 0.07 0.55 1 24.77
AVG (n=8) 0.913 0.07 0.53 19.89
SE 0.014 0.05 0.08 1.97

50% MVC DATA

Subject Shift Correlation Rest Avg. Average Reference Percent
Force (v) wave MVC
1 1.00 0.94 -0.01 1 .44 1 34.29
2 1 .00 0.95 028 1 .54 1 38.50
4 1 .00 0.96 0.24 1.79 1 55.94
6 1.00 0.92 0.25 0.43 1 30.71
7 1.00 0.96 0.10 0.38 1 34.55
10 1.00 0.91 0.23 0.97 1 23.10
11 1.00 0.65 0.11 0.17 1 17.00
12 1.00 0.96 0.10 0.91 1 40.99
AVG (n=8) 0.92 0.10 1.04 35.85
SE 0.03 0.06 0.17 3.38

MAXIMAL MVC DATA

Subject Shift Correlation Rest Avg. Average Reference Percent
Force (v) wave MVC
1 -1 .00 0.96 0.05 2.55 1 60.71
2 1 .00 0.94 -0.31 2.22 1 55.50
4 1 .00 0.95 0.26 2.64 1 82.50
6 1.00 0.93 0.38 0.53 2 37.86
7 0.00 0.91 0.10 0.30 1 27.27
10 1 .00 0.95 0.37 1 .26 1 30.00
1 1 0.00 0.92 0.01 0.64 1 64.00
12 0.00 0.96 0.12 1.19 1 53.60
AVG (n=8) 0.94 0.12 1 .64 54.63
SE 0.00 0.07 0.32 5.04

 

31

above 70%, while 3 subjects produced an average muscle force of 40% MVC or lower
during these first four cycles of the maximal MVC trial (Table 1). Similarly, the mean
forces during both the "25%" and "50%" exercise were somewhat less than the target
forces.

As expected, there was no substantial fatigue (i.e., decrease in mean force) over

the course of the 4 contractions during the 25% and 50% exercise. There was a

 

significant decrease in mean force between the first two vs. the last two contractions over
the course of the 8 contraction MAX exercise (Figure 11, p<0.003). However, even in
this case, the decrease (10.9 1 3.9%) was not as great as anticipated.
Region of fMRI Activation

In the subset of eight subjects, voxels within the primary motor cortex became
active with exercise in all three conditions as expected (e.g., Figure 12). Although other
regions such as the ipsilateral motor areas and cerebellum were active in some subjects,
this was not consistently observed in all subjects. In order to check that the alternating
"REST"-"WORK" cue did not itself result in activation, fMRI data was collected in one
subject who was instructed to watch the screen, but to perform no exercise. In this trial,
no activation was observed in any brain region including the primary motor areas. This
suggests that the observed activation during exercise is not dependent on the visual cues,

holding the hand grip, magnet noise, or other incidental factors.

32

Loss of force during maximal exercise condition
70

60- . l

50-

 

 

40-

 

‘ .
. l
30 d . i a
._ .. 02' g.) 7'} t
‘ b
j 5

Percent MVC (%)

20‘

10‘

 

 

 

 

 

: ' - so
0 T ' H

1

1st 2 cycles last 2 cycles

 

Figure 11: Loss of force from the first two cycles to the last two
cycles

Average data from the 100% MVC trial (N=8) from cycles 1 and 2 were averaged and
compared to the pooled data from cycles 7 and 8.

33

 

Mean Voxel Intensity values for the primary motor cortex

 

 

 

 

 

 

 

 

1060
—+— MAX MVC
~~-~o 50%MVC

1050 - —~— 25% MVC
1040 - V
I V

.5 J. ., J‘. 11 ,1 A if.

g 1030 - \ V 1w 1 1,

E ’0 (90 i ’0 "K i o H'09) Vv

— i.- ' :1. o m l ' '

92 1020 - 'me°o‘b° 09., “’5'" if? Vlz‘gai 0° 560°” °°°o

g i) 9; 9% b. ? ° 00 :' 3

1010 5 09 : Q. s 2 s 2,

‘ ‘5’- " ' 5 99 S ' 5 -

°' 0,631, 9.95 °=e 0&2 <.'> iii.
0 o 02,9 00 £130 6

1000 ~

990 1 r . . . i
0 50 100 150 200 250 300

Time (s)

Figure 12: Signal Intensity of voxels from the primary motor
area versus time for 3 different force levels

Data from MAX MVC represents the first half of the 510 second trial (N=8). The 25 and
50% trials are shown in their entirety (N=8).

34

Mean Voxel Intensity values during 100% MVC trial (n=8)
1070

 

10604

1050 -

 

. ‘ l
1040- .,«.l “ 1' l 1.. ‘

Voxel Intensity

 

1030 - > 1 1

1020 -

 

 

1 010 I I I I I
O 100 200 300 400 500 600

Time (s)

Figure 13: Mean intensity of voxels from the primary motor
area versus time for 100 % MVC

Data from MAX MVC represents the full 510 second trial (N=8).

35

 

FMRI response to each of the exercise conditions

Figures 12 and 13 summarize the mean intensity from the ROI selected as
described in Chapter 2 for each of the exercise conditions, and prior to motion correction.
Figure 12 shows the ROI intensity for the 25%, 50% and the first half of the MAX
exercise. Figure 13 shows the entire time course of the intensity variations during the
MAX exercise. There was some variability in baseline intensity between the various runs.
Within a trial, there was a consistent return to baseline during the rest periods. However,
that return to baseline was clearly slower than the return of force to baseline (Figure 12).
Presumably, this delayed recovery is due to a hemodynamic delay. As described in
Chapter 2, we accounted for this effect by fitting the results to reference waves with

broader on phases and shorter rest phases.

An interesting qualitative observation is the presence of a spike or peak at the
onset of each of the contractions in all of the exercise conditions (figures 12 and 13).
This peak appears more pronounced in the 50% and maximal MVC trials and appears to
coincide with the initial peak in force. Peaks at the end of many cycles are also observed
(e.g., Figure 12). All of these peaks are occuning when the hand is changing from one
state to another, i.e., either from resting state to an active muscle contraction or from an

active muscle contraction back to resting state.

Figures 14 and 15 show the intensity changes from the same ROI's after motion
correction using the AFN I plugins. Although in some minor respects the results are
modified (e.g., the drift of the last cycle in Figure 13 is corrected after motion correction),

the basic results are not significantly altered.

36

Motion corrected voxel intensities

 

 

 

 

 

 

 

 

1110
—0- 100%MVC
-----o 50%MVC I
1100 _‘ —"— 25% MVC l|
fw‘ '1 all
It I
b I 1 IV 11
@1090“ lie at t 1* llwn‘i ii l
93 ‘ I <90 1 [I 1‘ 1" a life LI
5 A“ l A db . 99°.h p c
'76 «1° . . . 4.. 91° \Uo‘i.
E, 1080 4 5 °. ‘ .2 550° é V
.°. . p . - ,9 ..
. .-... .i ’ O 0
10701 6 ° ° 8’5 . .. ‘ 80-";
1060 . . . . i .
0 50 100 150 200 250
Time (s)

Figure 14: Motion corrected voxel intensity from the primary
motor area for 3 exercise trials

Data from the 25%, 50% and Maximal MVC trials (N=8) after motion correction

techniques are compared. Data from the maximal MVC trial includes only the first 270
seconds of that trial.

37

 

1100

 

 

 

 

 

 

 

 

 

 

 

 

 

Motion corrected VI values during 100% MVC trial
1090 -
1
g
(D
C
Q I
5 1030 ~ Hi
i.) 0
g 0 0 0
> 0 0
ll 1 l
1060 I I I l l f
0 100 200 300 400 500 600
Time (s)

Figure 15: Motion corrected signal intensity of voxels from the
primary motor area during the maximal MVC trial

Data from the 100% MVC trial (N=8) after motion correction techniques were applied.
The abnormally high peaks occurring in figure 13 have been corrected. Eight distinct
cycles of signal activation are visible.

38

Table 2: Summary of raw voxel intensity data
25% RAW v0xer mreusn'v DATA

 

 

 

 

 

 

 

 

 

 

 

Subject Shift Correlation Rest Avg. Intensity increase Reference Percent
wave Chang_e_
1 1 0.88 1058 17.2 4 1 .628
2 1 0.75 840 17.9 6 2.1
4 5 0.54 1030 9.48 2 0.92
6 1 0.53 1175 4.9 5 0.41
7 1 0.64 1 132 5 5 0.44
10 1 0.57 932 9.6 5 1 .03
11 2 0.39 1064 5.23 7 0.491
12 1 0.83 946 12.07 4 1.27
AVG (n=8) 0.64 1022 10.17 1 .04
SE 0.0591 39.10 1 .85 0.21
50% - RAW VOXEL INTENSITY
Subject Shift Correlation Rest Avg. Intensity increase Reference Percent
wave Chang_e_
1 1 0.86 1070.00 18.20 4 1.70
2 1 0.88 849.00 26.00 5 3.10
4 1 0.82 1033.00 17.70 5 1.70
6 1 0.80 1 161.00 8.70 4 0.75
7 1 0.71 1089.00 8.50 4 0.78
10 1 0.79 926.00 16.40 6 1.77
1 1 2 0.38 1036.00 5.59 3 0.54
12 0 0.75 881.00 14.55 3 1.65
AVG (n=8) 0.75 1005.63 14.46 1.50
SE 0.06 38.57 2.35 0.29
100% - RAW VOXEL INTENSITY
Subject Shift Correlation Rest Avg. Intensity increase Reference Percent
wave Change—
1 0 0.59 1032.00 15.50 3 1.50
2 -7 0.70 855.00 24.00 2 2.80
4 1 0.85 1033.00 16.60 5 1.60
6 1 0.74 1 188.00 9.90 4 0.83
7 1 0.66 1142.00 9.50 7 0.83
10 1 0.86 925.00 23.10 5 2.49
11 1 0.67 1114.00 14.10 6 1.27
12 1 0.84 921.00 23.40 4 2.60
AVG (n=8) 0.74 1026.25 17.01 1.74
SE 0.04 41 .83 2.09 0.28

 

Table 3: Summary of motion corrected voxel intensity data
25% - MOTION CORRECTED VOXEL INTENSITY
Subject Shift Correlation Rest Avg. Intensity increase Reference Percent

 

 

wave Change
1 1 0.85 1132 15.7 4 1.38
2 1 0.73 916 14.5 4 1 .58
4 5 0.6 1013 15.01 1 1.48
6 0 0.17 1336 1 .28 6 0.1
7 1 0.57 1156 7.52 4 0.65
10 7 0.48 947 1 1 .89 1 1 .26
11 7 0.37 1151 5.86 1 0.51
12 1 0.55 1008 10.64 2 1.06
AVG (n=8) 0.54 1082 10.30 1.00
SE 0.0739 48.78 1.79 0.187

 

 

50% - MOTION CORRECTED VOXEL INTENSITY

Subject Shift Correlation Rest Avg. Intensity increase Reference Percent
wave Change

 

 

1 1 0.87 1 142 18.46 3 1 .62

2 1 0.89 926 20.41 4 2.2
4 1 0.84 1048 27.8 6 2.65

6 -7 0.23 1294 2.53 1 0.2
7 1 0.77 1 164 13.84 3 1 .19
10 6 0.75 971 17.68 1 1.82

1 1 6 0.19 1030 3.07 1 0.3
12 0 0.75 977 13 4 1 .34
AVG (n=8) 0.66 1069 14.599 1.42
SE 0.10 43.39 3.03 0.30

 

100% - MOTION CORRECTED VOXEL INTENSITY
Subject Shift Correlation Rest Avg. Intensity increase Reference Percent

 

 

 

wave Change

1 0 0.78 1124 17.68 3 1.57
2 -7 0.73 926 20.30 2 2.19
4 1 0.89 1014 29.30 5 2.89
6 7 0.48 1286 7.10 2 0.55
7 1 0.59 1 108 6.52 4 0.59
10 1 0.78 945 22.72 6 2.40
11 4 0.37 1 146 7.16 3 0.62
12 1 0.73 1002 16.43 4 1.64
AVG (n=8) 0.67 1068.88 15.90 1 .56
SE 0.06 42.45 2.96 0.32

 

40

The detailed results of the ROI vs. reference wave correlation analysis are
summarized in Tables 2 (before motion correction) and 3 (after motion correction). The
column labeled “Shift” in the Tables represents the number of points that the reference

was shifted in order to obtain the best fit with the fMRI data. Positive numbers reflect

shifts to the right, while negative numbers reflect shifts to the left. Figure 9 for example,
has a rightward shift of 1, corresponding to a delay of 3 seconds. The fact that in almost
all cases the shift was positive is consistent with the expected hemodynamic delay
between neural activity and the BOLD response. The "wave" in the Tables refers to
which of the 7 reference waves (Figure 8) gave the best fit. The fact that in most cases the
broader on-phase references gave the best fit confirms the qualitative impression that the

hemodynamic response persisted into the rest periods.

There was as significant increase in the percent change in ROI intensity for the
50% and the first 4 contractions of the MAX exercise compared to the 25% exercise
(p<0.025 for 50% vs. 25%, paired t—test; p<0.007 for MAX vs. 25%). This conclusion
was essentially unaltered by the motion correction (Figure 16). However, the MAX trial
did not statistically differ from the 50% MVC trial (P > 0.15 for both non-motion
corrected and motion corrected data). The variance within each of the groups was not
equal using the Levine test of homogeneity of variance (P < 0.05). The unequal variance
is not too great a concern as the crucial test is the paired comparison between exercises.

Figure 17 shows the relationship between the ROI intensity changes and the force
developed during the first four contractions of the three exercises for the individual
subjects. The lines are simple linear regressions for each subject. Because the variations

in both force and ROI intensity varied widely between different subjects, a simple

41

Percent Change in Voxel intensity -
(Both Pre and Post Motion correction)

 

313*

 

- Pre - (8 subjects)
a Post

 

 

 

Percent Change (%)

 

 

 

 

 

 

 

 

0 1 0 20 30 4O 50 60 70
Percent MVC

Figure 16: Bar graph showing percent change in voxel
intensity in raw and motion corrected data

Black bars represent pre motion corrected data. Grey bars represent motion-corrected
data. * p<0.025 for 50% vs. 25%, ** p<0.007 for MAX vs. 25%. No difference exists
between pre motion correction data and post motion corrected data.

42

8 subject - Individual regression

 

 

 

 

 

 

 

4
3..
__ [II
>
.E O O
%2- A
as J O (7
6 1 <7 ”T
i“
a) V V O
0. 1 . o
I
O...
0 20 40 60 100
PercentMVC

Figure 17: Simple linear regressions of individual subjects

Each individual correlation shows an upward trend as the relative force level increases so

too does the percent change in voxel intensity.

43

regression of data from all the subjects was not significant. While there is a general trend
for increased fMRI response with increased force, this trend may be obscured by the fact
that the force during the MAX exercise was not greatly different than during the 50%

exercise.

Effects of fatigue on signal intensity

In order to induce muscle fatigue, the maximal MVC trial was twice as long as the
other two trials. However, as discussed above, the decrease in force during this exercise
was only (10.9 1 3.9%, p<0.003, see Figure 11). Figure 18 shows the ROI intensity
changes during the first two contractions vs. the last two contractions of the MAX
exercise. No significant difference in signal intensity was observed for either the non-
motion corrected time series (P> 0.05 ) and the motion corrected time series (P>0.05).
The non-motion corrected time series showed a high degree of variability between
subjects and was noticeably reduced by motion correction. In any case, neither analysis

provides evidence that the fMRI response was altered by fatigue.

Percent change in ROI voxel Intensity
during maximal exercise condition

 

 

 

 

 

 

 

 

 

 

 

5
A — pre motion correction
°\° E223 post motion correction
z" 4 " HF
75
C
0:
Que
.E
92 3~
0
>
.E
8.
r: 2 '
m
.C
O
‘8
e 1 ‘
o
a.
0

   

1st 2 cycles last 2 cycles

Figure 18: Comparison of first 2 cycles to the last 2 cycles of
the raw and motion corrected voxel intensity data

There is no significant difference between the first 2 cycles and the final 2 cycles in either
the pre motion corrected or motion corrected data.

45

Chapter 4: Discussion and Future Directions

The main results of this study are, first, that increased force of isometric muscle
contraction is associated with increased intensity of the BOLD response in the motor
cortex, and second, that muscle fatigue is not accompanied by a significant change in the
BOLD response during isometric contraction.

The first aim of the study involved varying levels of muscle force to determine if
this would have an effect on the intensity of the BOLD response in the primary motor
cortex. We found an increase in ROI intensity which was significant between 25% MVC
and 50% or "MAX" conditions. There was no significant difference between the 50%
and MAX conditions, but this result may simply reflect the fact that our subjects did not
actually develop their maximum MVC during that trial. Because our data only extends
from about 20% to 55 % of MVC (e.g, Figure 16), we have not fully characterized the
dependence of the fMRI response on force. However, based on our results over this
limited range, it appears that the relationship between isometric force and fMRI response
is not linear. There are at least two possible explanations for the apparent non-linear
response. First, it may be that the neural activity of the R01 is accurately measured by
MRI, and that activity is not linear with force. For example, it could be that only a
fraction of the neurons in the R01 are involved in force generation, and that the ROI
includes other neurons which are involved in such things as initiation of force, feedback
regulation, regulation of the rate of force development and relaxation, etc. The activity of

these other neurons may not depend on the absolute force developed during the

46

maintained contractions. Second, it may be that the fMRI response is not an accurate
measure of the average neuron activity in the ROI. For example, the BOLD response may
approach saturation at less than the maximum force generation. Alternatively, it may be
that metabolism in the glial cells contributes a substantial component to the BOLD effect,
and that glial metabolism is not proportional to the underlying neural activity.

Although the relationship was not linear, this is the first study to report a
significant increase in BOLD intensity vs. force of muscle contraction. These findings are
not consistent with the earlier results of Ludman et al., who reported no effect of force on
MRI response during a dynamic weight-lifting exercise (Ludman et a1. 1996). However,
it should be noted that the intensity of the motor cortex response reported in Ludman et
al. was 2-5%, whereas the greatest mean change observed in this study was less than 2%.
Therefore, the overall response to dynamic contractions is clearly greater than to
isometric contractions. This difference between dynamic vs. isometric contractions was
previously reported by Thickbroom et a1. (Thickbroom et a1. 1999). In fact, in their first
paper, Thickbroom et a1. did report a trend toward increased response with increased
force during isometric contractions, but the trend did not reach statistical significance
(Thickbroom et a1. 1998).

The second aim of this study was to examine if fatigue is accompanied by a
significant change in the intensity of the response in the activated region of the motor
cortex. We found no significant difference between the intensity of the response near the
start vs. near the end of the MAX exercise (Figure 18). However, the extent of fatigue in
this study was not dramatic, largely because the subjects did not begin contracting at their

measured MVC. Therefore, the failure to observe any change in the MRI response is not

47

conclusive. It does appear that the marginal fatigue observed in this study is not
accompanied by a measurable change in MRI response. Therefore, based on this limited
result, there is no evidence for a "central component" in the fatigue. Of course, the failure
to detect a change in response with fatigue in this experiment does not mean it could not
be observed under different conditions. In fact, our protocol may not be ideal for causing
central fatigue. As mentioned in Chapter One, central fatigue is thought to be more
important at low levels of force (20-25%) over a longer duration of muscle contraction
(20 minutes or more, vs. 8.5 min in this study). Unfortunately, a study of long duration
exercise would raise many other technical problems, for example, increased chance of
motion artifacts and instrumentation errors.

It may be that presenting the subjects with visual feedback of their force would
have improved subject compliance during the MAX exercise, and resulted in greater
fatigue during that exercise. For example, in the studies by Thickbroom et al., subjects
could see the force developed on a sphygmomanometer. However, we chose not to
display the force record to the subjects during the experiments for two reasons. First, in
our early pilot studies, we did allow visual feedback during the exercise trials. This
allowed for maintenance of the target force level, but subjects tended to neglect the need
to prevent head motion, especially at the higher forces. Subjects demonstrated accuracy
at each of the lower levels of force, i.e., levels below 60% MVC. However, at higher
forces, the subjects began to use arm and shoulder muscles in order to achieve the target
force. This always resulted in some sort of head motion, and excessive motion artifacts

in the functional images. Second, as discussed in Chapter 1, we were concerned that the

48

feedback would involve an additional cognitive activity, which might obscure activity
dependent on the force per se.

Despite our efforts to limit head motion, motion artifacts were still a problem in
this study, and resulted in rejection of data from three of the subjects. A number of
techniques have been used to reduce head motion, including clamping the head in place
to restrict movement and the use of bite bars. Head clamping hasn’t been successful for
longer scans as pressure points on the head cause discomfort and subject motion (Zeffiro
1996). Bite bars can be successful for reducing head motion but are uncomfortable, and
anecdotal reports suggest they increase the sensation of claustrophobia in the magnet.
Because our subjects volunteered without any compensation, the use of the bite bar was
rejected for this study. Another strategy to prevent head motion involves behavioral
feedback. This involves training the subject to remain motionless during scans and is
effective in reducing head motion (Zeffiro 1996). Unfortunately, in this experiment
extensive training was not practical, again, because the subjects were asked to volunteer
without any compensation.

In addition to the main results, we observe spikes in the intensity of the ROI at the
beginning and end of the "on" cycles of the exercise. These spikes coincided with an
active effort from the subject to either generate muscle force or to relax the hand. This
suggests that changing from one static level of muscle activation to another require an
increase in neural activity in the selected ROI (Toma et al. 1999). The observed spikes
could be related to the more robust fMRI response that occurs during dynamic exercise
(Thickbroom et al., 1999). Even though our isometric handgrip exercise did not allow

any joint flexion, a sensory and proprioceptive component is present and could be

49

enhanced at the beginning and ending of the contractions. At the onset of an isometric
contraction, some shortening of the muscle fibers must occur and as muscle tension rises,
sensory receptors will be activated in the muscles, tendons and in the joints of the hand.
This milieu of sensory feedback likely isn’t as pronounced as in a dynamic exercise, but
could still be sufficient to cause the spikes.

Based on the experience gained in this study, future studies could be improved in
some respects. For example, it would be helpful to have subjects practice the entire
protocol out of the magnet prior to the actual imaging. In practice sessions, visual
feedback would be included so that the subjects would be aware of the tendency to
decrease force over the course of the 30 s contractions. Practice would also provide an
opportunity for the investigator to coach the subjects not to recruit neck and shoulder
muscles during the exercise, and to remind subjects to lie still during the rest periods.
Finally, subjects could be better trained in the use of the handgrip device during practice
sessions. This is important, because if the subject repositions his/her hand during the
experiment, the lever arm of the device changes, and the effective MVC can change. The
practice would have to be held on separate days from the actual fMRI sessions to allow
for recovery from muscle fatigue. A 2-3 day commitment from subjects would likely be
necessary. Development of a more comfortable handgrip would also be worthwhile.
Some subjects, especially those with smaller hands, found the grip to be slippery, and had
difficulty maintaining a secure grip. A device that fixed the hand into position, for
example a glove-like device, would be an improvement.

Another improvement would be to acquire sagital slices during the fMRI

acquisitions rather than axial slices. The most severe motion artifacts encountered in this

50

study appeared to be due to movement along the superior-inferior axis, rather than to
rotation or translation in the axial plane. Because we used axial slices, the spatial
resolution in this study is lowest in the superior-inferior direction, making motion
correction more difficult in this direction. Although the motion correction algorithms did
improve the quality of the data, particularly near the end of the MAX exercise, these
algorithms could not rescue three of the 13 data sets.

A logical continuation of this study would be to repeat the same basic protocol,
but switch from one force level to another force level rather than returning to rest during
each cycle (e.g., using 25% MVC as the resting state and 50% MVC as the active state).
This would check both the specificity of the intensity changes to force level, and also
enable exploration of the cause of the spikes observed at the onset and end of the
contractions. Another useful variation would be to simply instruct the subjects to change
the force at regular intervals, but give no detailed pattern of target forces. In such a
protocol, the force record (appropriately shifted as necessary) could be used as the
reference wave for cross-correlation statistics.

In summary, there were two major findings of this study. First, the results show
that there is a significant effect of isometric force on the fMRI response in the
contralateral motor cortex. Second, the results provide no evidence that the fMRI
response in the motor cortex is altered by muscle fatigue. However, further studies will
be needed to demonstrate the generality of these results to other exercise and fatigue

conditions.

51

References

Adams, G.R., M.J. Fisher, RA. Meyer. (1991) Hypercapnic acidosis and increased
H2P04' concentration do not decrease force in cat skeletal muscle. Am J Physiol 260(4 pt
1):C805-12.

Bandettini, P. A., E. C. Wong, et al. (1992). “Time course EPI of human brain function
during task activation.” Magn Reson Med 25(2): 390-7.

Belanger, A. Y. and A. J. McComas (1981). Extent of motor unit activation during effort:
J Appl Physiol 1131-1135.

Bellemare, F., F. Furbrush, B. Bigland-Ritchie (1984). Localization of human diaphragm
fatigue. Fed prod Abst, 42:530, 1984.

Bigland-Ritchie, B. and J. J. Woods (1984). Changes in muscle contractile properties and
neural control during human muscular fatigue: Muscle Nerve 691-699.

Bigland-Ritchie, B., F. Furbush and J. J. Woods (1986). Fatigue of intermittent
submaximal voluntary contractions: central and peripheral factors: Journal of Applied
Physiology 61:421-429.

Blamire, A. M., S. Ogawa, et al. (1992). Dynamic mapping of the human visual cortex by
high-speed magnetic resonance imaging. Proc Natl Acad Sci U S A 89(22): 11069-73.

Boroojerdi, B., H. Foltys, et al. (1999). “Localization of the motor hand area using
transcranial magnetic stimulation and functional magnetic resonance imaging.” Q1113
Neurophysiol 110(4): 699-704.

Brust, J. (1991). Cerebral circulation: Stroke. Principles of Neural Science. E. Kandel,
J.H. Schwartz, T.M. Jesse]. New York, Appleton and Lange: 1041-1049.

Cady, E.B., D.A. Jones, J. Lynn, D.J. Newham, (1989). Changes in force and intracellular
metabolites during fatigue of human skeletal muscle. J Physiol, 418: 311-25.

Chen, R. (2000). “Studies of human motor physiology with transcranial magnetic
stimulation.” Muscle Nerve 999(9): 826-832.

Cote L., M. D. C. (1991). The Basal Ganglia, chapter 42. Principles of Neural Science,
Md; J. H. S. E.R. Kandel, T.M. Jesse] (eds). New York, Appleton and Lange: 647-
654.

Cox, R. W. (1996). AFNI: software for analysis and visualization of functional magnetic
resonance neuroimages: Computers and Biomedical research, 162-173.

52

Deecke, L., P. Scheid, et al. (1969). “Distribution of readiness potential, pre-motion
positivity, and motor potential of the human cerebral cortex preceding voluntary finger
movements.” Exp Brain Res 7(2): 158-68.

Dettmers C., G.R. Fink, R.N. Lemon, K.M. Stephan. R.E. Passingham, D. Silbersweig,
A. Holmes, M.C. Ridding, D.J. Brooks, R.S.J. Frackowiak (1995). Relation between
cerebral activity and force in the motor areas of the human brain. J Neumphysiol, 74:802-
815.

Ellerman J, M. G., K Hendrich, R Hinke, X Hu, SG Kim, R Menon, H Merkle, S Ogawa,
K Ugurbil (1994). Functional imaging of the brain by nuclear magnetic resonance. NMR
in Piusiology and Biomedicine. G. RJ. New York, Academic Press: 137-150.

Enoka, R. M. and D. G. Stuart (1992). Neurobiology of muscle fatigue: J Appl Physiol
163 1-1648.

Enoka, R.M., (1996) Eccentric contractions require unique activation strategies by the
nervous system. J Appl Physiol, 81(6): 2339-46.

Farrar, T., DE Becker (1971). PulsrL and Fourier Transform NMR. New York, Academic
Press.

Fitts, RH. (1994). Cellular mechanisms of muscle fatigue. Physiological Reviews,
74(1):49-94.

Fox, P.T., M.A. Mintun, M.E. Raichle, FM. Miezin, J .M. Allrnan, D.C. Van Essen.
(1986) Mapping human visual cortex with positron emission tomography. Narure,
323(6091):806-9.

Ganong, W. (1991). Ciculation through special regions. Review of Medical Physiology,
15th ed. Norwalk, CT, Appleton and Lange: 570-572.

Ghez, C. (1991). The cerecellum. Principles of Neural Science, 3rd Ed... J. H. S.
Kandel, T.M. Jesse] (eds). New York, Appleton and Lange: 618-641.

Grubb, R. L., In, M. E. Raichle, et al. (1974). “The effects of changes in PaCO2 on
cerebral blood volume, blood flow, and vascular mean transit time.” Stroke 5(5): 630-9.

Gyngell, M. L., C. Bock, et al. (1996). “Variation of functional MRI signal in response to
frequency of somatosensory stimulation in alpha-chloralose anesthetized rats.” Magn
Reson Med 36(1): 13-5.

Hales, J. P. and S. C. Gandevia (1988). Assessment of maximal voluntary contraction

with twitch interpolation: an instrument to measure twitch responses: J Neurosci
methods 97-102.

53

Homak, J. P. (2000). The basics of MRI. http://www.cis.rit.edu/htbooks/mrilinside.htm.
2000.

Ilmoniemi, R. J ., J. Ruohonen, et al. (1999). “Transcranial magnetic stimulation--a new
tool for functional imaging of the brain.” Crit Rev Biomed Eng 27(3-5): 241-84.

Kandel ER, S. S., Schwartz J H. (1991). Synaptic Transmission. Principles of Neural
Science, 3rd Ed. J. H. S. E.R. Kandel, T.M. Jesse]. New York, Appleton and Lange:
123-134.

Kelly, J. P. (1991). The neural basis of perception and movement. Principles of Neural
Science, 3rd Ed. I. H. S. E.R. Kandel, T.M. Jesse] (eds). New York, Appleton and
Lange: 293-295.

Koechlin, E., G. Corrado, et a]. (2000). “Dissociating the role of the medial and lateral
anterior prefrontal cortex in human planning.” Proc Natl Acad Sci U S A 97(13): 7651-6.

Lim, S. H., D. S. Dinner, et al. (1994). “Functional anatomy of the human supplementary
sensorimotor area: results of extraoperative electrical stimulation.” Electroencephalogr

Clin Neurophysiol 91(3): 179-93.

Ludman, C. N ., T. G. Cooper, et al. (1996). “Force of voluntary exercise does not affect
sensorimotor cortex activation as detected by functional MRI at 1.5 T.” NMR Biomed
9(5): 228-32.

Martin, J. H., J.C. Brust, S. Hilal (1991). Imaging the Living Brain. Principles of Neural
Science, 3rd Ed. E. R. Kandel, J .H. Schwartz, T.M. Jesse]. New York, Appleton and
Lange: 309-324.

McKenzie, D.K., B. Bigland-Ritchie, R.B. Gonnan, S.C. Gandevia. (1992). Central and
peripheral fatigue of human diaphragm and limb muscles assessed by twitch
interpolation. J Physiol, 454: 167-76.

Merton, P. A. (1954) Voluntary strength and fatigue, J. Physiol. London, 123:553-564.

Merton, P. A. and H. B. Morton (1980). “Stimulation of the cerebral cortex in the intact
human subject.” Nature 285(5762): 227.

Ogawa, S., T. M. Lee, et a]. (1990). Brain magnetic resonance imaging with contrast
dependent on blood oxygenation: Proc Natl Acad Sci 9868-9872.

Ogawa, S., R. S. Menon, et al. (1998). “On the characteristics of functional magnetic
resonance imaging of the brain.” Annu Rev Biophvs Biomol Stru_ct 27: 447-74.

54

Ogawa, S., R. S. Menon, et al. (1993). “Functional brain mapping by blood oxygenation
level-dependent contrast magnetic resonance imaging. A comparison of signal
characteristics with a biophysical model.” Biophys J 64(3): 803-12.

Olesen, J. (1971). “Contralateral focal increase of cerebral blood flow in man during arm
work.” Brain 94(4): 635-46.

Porter, R. J. W. (1981). Human Muscle fatigue: physiological mechanisms. London,
Pitrnan Medical.

Raichle, M. E., R. L. Grubb, Jr., et al. (1976). “Correlation between regional cerebral
blood flow and oxidative metabolism. In vivo studies in man.” Arch Neurol 33(8): 523-6.

Ramsey, N. F., B. S. Kirkby, et al. (1996). “Functional mapping of human sensorimotor
cortex with 3D BOLD fMRI correlates highly with H2(15)O PET rCBF.” J Cereb Blood
Flow Metab 16(5): 755-64.

Rao, S.M., J.R. Binder, P.A. Bandettini, T.A. Hammeke, F.Z. Yetkin, A. Jesmanowicz,
L.M. Lisk, G.L. Morris, W.M. Mueller, L.D. Estkowski, E.C. Wong, V.M. Haughton,
J .S. Hyde. (1993) Functional magnetic resonance imaging of complex human
movements. Neurology, 43, 231 1-2318

Rapoport, S. I. (1976). Bllod-Brain Bgrrier in Physiology and Medicine. New York,
Paven Press.

Roberts, T. P. and H. A. Rowley (1997). “Mapping of the sensorimotor cortex: functional
MR and magnetic source imaging.” AJNR Am J Neuroradiol 18(5): 871-80.

Roland, E. and B. Larsen (1976). “Focal increase of cerebral blood flow during
stereognostic testing in man.” Arch Neurol 33(8): 551-8.

Roland, P. E., B. Larsen, et al. (1980). “Supplementary motor area and other cortical
areas in organization of voluntary movements in man.” J Neurophysiol 43(1): 118-36.

Rothwell, J. C. (1997). “Techniques and mechanisms of action of transcranial stimulation
of the human motor cortex.” J Negrosci Methods 74(2): 113-22.

Savoy, R.L. (2001) History and future directions of human brain mapping and functional
neuroimaging. Acta Psychol, 107(1-3):9—42.

Schwartz, J ., Ed. (1991). Synaptic Vesicles. Principles of Neural Science, 3rd Ed., New
York, Appleton and Lange.

Shaw, D. (1976). Relaxation. Fourier Transform N.M.R. Sfltroscopy. S. D. New York,
Elsevier Scientific Publishing Company: 297-298.

55

Schleicher A, A. K., Geyer S, Kowalski T, Schorrnann T, Palomero-Gallagher N, Zilles
K (2000). “A stereological approach to human cortical architecture: identification and
delineation of cortical areas.” J Chem Neuroanat 20(1): 30-47.

Thickbroom, G., B. A. Phillips, et al. (1998). Isometric force-related activity in
sensorimotor cortex measured with functional MRI: Exp Brain Res, 121:59-64.

Thickbroom, G. W., B. A. Phillips, et al. (1999). “Differences in functional magnetic
resonance imaging of sensorimotor cortex during static and dynamic finger flexion.” Qtp
Brain Res 126(3): 431-8.

Thickbroom, G. W., M. L. Bymes, et al. (2000). “The role of the supplementary motor
area in externally timed movement: the influence of predictability of movement timing.”
Brain Res 874(2): 233-241.

Thulbom, K. R. (1999). Visual feedback to stabilize head position for MRI: 1039-1043.
Yoshizawa, T., T. Nose, et a]. (1996). “Functional magnetic resonance imaging of motor
activation in the human cervical spinal cord.” Neuroimage 4(3 Pt 1): 174-82.

Toma, K., M. Honda, T. Hanakawa, T. Okada, H. Fukuyama, A. Ikeda, S. Nishizawa, J.
Konishi, H. Shibasaki. (1999) Activities of the primary and supplementary motor areas
increase in preparation and execution of voluntary muscle relaxation: An event-related
fMRI study. J Neurosci, 19(9): 3527-34.

Yetkin, F. 2., W. M. Mueller, et al. (1995). “Functional magnetic resonance imaging
mapping of the sensorimotor cortex with tactile stimulation.” Neurosurgeg 36(5): 921-5.

Zeffiro, T. (1996) Clinical functional image analysis: artifact detection and reduction.
Neuroimage, 4(3 pt 3): S95-100.

56

   

'lHallllzlllllllljllgl'glljlljll '