' IV-P}
. '0‘ ”3:“

'4

L~

.4!
I’\

I

‘ 9.5;; 1.

L
“-3

J.

.vu
rm

.. 7;
iaz'
1.,

_ r

‘pg- .n‘ 14‘3"

. 'W“ ‘ .,-—
.z- ‘.

.
. 1. r.
,3,

{m'

7 ‘7 :91: .
' 'F‘J. "
2 “‘2

, ,
.. 4.45] ‘
4 :v‘ ;4

:5.
A

‘35
.l‘
,1 r 1.1%.} are}:

.....-r
«.1 1-,

2'
443367

N. v ,\‘, -
.. . WW
:42“: ric‘ 21*“ ..~‘ ‘
:5" he
«14);. ~v.‘ ~-:n_§a.,
vu‘fl“ I.
4. 4; . 11w
1: z; .
-.r‘4;a l. ,, $2512? 9' ,
.. . k

m, V
n

r

4 ,.
n:

”it”
4 r (v (-
1». 3,3517}.

y?{$
4 ~'

(if?
if

"I"
‘3

.4
“I

_,“
I.

.11! -\-,.
“tn Ln:

..
.~ .;!.
‘ r

4,; r

1.x},

)4‘ '-
>1 ' {i
‘.|.4,2.

w v
’1”?!
A" -'

L.
.
,,,
.‘,z.-r”e.""
; u.
of
¢ l’
41.," .. . x 1,47...
_,.., .3.“ :51 .
‘4‘ f '.
.4
I I ' v
-u n ’I
' 1

r"
_-,.

- -4
‘ , «jar-A , '- 1‘.
4 - ‘ , , ‘ .1 .' l I - ' ‘

nun". ‘ M r '3» A. - ‘ ‘44-» hi . J5.

, \ A ,._. . _ . _,f .2}; 4 ' 1mm

. "a , ~.,. 4;. ""7. u -. ,r. (inf—w.-
uxl-wr; ‘ l W ‘ "tr

0;

'4‘
;

 

4'.
.

 

1w

J MICH'GM STATE WNEHSWY LIBRARIES 2 8 5’9 3—2/2 X,
lllllllllllllllHllllllllllllllllllllllllllllHllllllllllllll

: 3 1293 00788 4046

; USERARY

l

:t-l Q ' “ ' I
585 ngclzsgan $3.1,

; University 3

L—

This is to certify that the
thesis entitled

NON-INVASIVE TEMPERATURE MEASUREMENT
BY ULTRASONIC TECHNIQUES

presented by
Di Ye

has been accepted towards fulfillment
of the requirements for

Mr S. degree in__E|echjggl Engineering

7

Major p essor

Date June 26, 1989

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.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

, , ]
MSU Is An Affirmative Action/Equal Opportunity Institution
cAcIrchma-pd

 

NON -INVASIVE TEMPERATURE MEASUREMENT

BY ULTRASONIC TECHNIQUES

By

Di Ye

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of
MASTER OF SCIENCE
Department of Electrical Engineering and Systems Science

1989

(90501;!)

ABSTRACT

NON-INVASIVE TEMPERATURE MEASUREMENT

BY ULTRASONIC TECHNIQUES
By
Di Ye

Non-invasive temperature measurement utilizing a micro-computer
based ultrasonic imaging system for in- vitro tissue studies has been investi-
gated. This work is based on the theory that the velocity of ultrasound in a
medium is a function of temperature, and on previous experimental results
of temperature dependent sensing in biological tissues. The work reported
here represents a special technique developed recently to monitor the rela-
tive temperature change in tissues in a simulated hyperthermia treatment
situation. The technique involves wave pattern matching to detect the echo
phase change due to temperature variation in the tissue. A 40 MHz sampling
rate data aquisition system is used to obtain high resolution in velocity
change so that a 0.20C temperature change may be identified. Temperature
profile imaging of an experimental model has been obtained. The techniques

used and experimental results obtained are presented.

Acknowledgements

I wish to thank Dr. Bong Ho, my advisor, for his timely guidance and
support. I wish to thank Dr. H. Roland Zapp and Dr. R. O. Barr for their
suggestions relevant to this research.

I wish to thank Mr. Nai Hsien Wang and Mr. Mo Zhang for their aid in
the use of the computer software developed in the Ultrasound Research
Laboratory and in assistance with equipment operation. I would like to thank

my parents and my wife for their constant help and encouragement.

iii

Table of Contents

List of Tables ............................................................................... vi
List of Figure ............................................................................... vii
Chapter 1 Introduction ................................................................. 1
Chapter 2 Theoretical Considerations ......................................... 5
2.1 Review of Ultrasound Principles .................................... 5
2.2 Temperature Dependence of Ultrasonic
Propagation ...................................................................... 9
Chapter 3 Acoustic Considerations for Temperature
Monitoring ........................................................................ 18
3.1 Ideal Step-Function Temperature
Distribution Model ........................................................... 18
3.2 Effect of Realistic Temperature Distribution ................. 21

3.3 Effect of the Nonlinear Relationship

Between Temperature and Velocity ................................ 25
Chapter 4 Velocity Measurement Techniques ............................ 30
4.1 Review of Temperature Measurement ........................... 30
4.1.1 Interferometric Techniques .......................................... 30
4.1.2 Pulse-echo Overlap Technique .................................... 31
4.1.3 Phase-slope Method ..................................................... 32
4.1.4 Cross-correlation Method ............................................ 36
4.1.5 Acoustic Phase Shift Technique .................................. 39
4.1.6 Other Techniques ......................................................... 43

iv

4.2 Temperature Measurement Technique

in This System ............................................................... 47

Chapter 5 System Configuration and Operation ......................... 49
5.1 Hardware Description ..................................................... 49
5.1.1 Non-intrusive Temperature Measurement System ...... 49
5.1.2 Intrusive Temperature Measurement System .............. 55

5.2 Software Description ...................................................... 58

5.3 System Operation ............................................................ 63
Chapter 6 Experiment Design and Procedures ............................ 65

6.1 Experiment 1: The Ultrasound Velocity Change

of Oil due to Temperature Variation ............................... 65
6.2 Experiment 2: Temperature Profile in Water ................. 72
6.3 Experiment 3: Temperature Profile Probing .
of a Layered Model .......................................................... 79
6.4 Experiment 4: Design Considerations and
Procedural Aspects for Biomedical Tissue Studies ......... 86
Chapter 7 Analysis of Experiment Results .................................. 94
7.1 Analysis and Discussion ................................................. 94
7.1.1 Accuracy and Time Resolution ................................... 94
7.1.2 Medical Applications ................................................... 98
7.2 Conclusion and Future Study .......................................... 101
Bibliography ................................................................................ 103

List of Tables

Table 2.1.1 Approximate values of ultrasonic velocities
of various media ........................................................ 8

Table 7.1.1 Change in ultrasonic velocity as a function
of temperature elevation and path length .................. 97

vi

List of Figures

Figure 2.2.1 Temperature variation of the velocity
of ultrasound in water ............................................... 14

Figure 2.2.2 Speed of sound vs. temperature in-vivo runs ......... 15

Figure 2.2.3 Variation of velocity of sound with temperature
in various tissues ....................................................... 16

Figure 2.2.4 Variation of velocity of sound with temperature
in breast tissues ......................................................... 17

Figure 3.1.1 The step-ftmction temperature profile .................... 20

Figure 3.2.1 Temperature - Duration threshold for
histological damage .................................................. 24

Figure 3.3.1 (a and b) Illustration of sound velocity as a
function of temperature across a non-uniform

temperature profile .................................................... 28

Figure 3.3.1 (c and (1) As above ................................................... 29

Figure 4.1.1 The phase - slope method ....................................... 34

Figure 4.1.2 The phase - slope method ....................................... 35

' Figure 4.1.3 Cross - correlation method ..................................... 38
Figure 4.1.4 The system operation for phase shift

technique ................................................................... 40

Figure 4.1.5 Phase shift technique .............................................. 41

Figure 4.1.6 Acoustic phase shift with temperature ................... 42

Figure 4.1.7.a Ultrasound computerized tomography system ..... 45

Figure 4.1.7.b Data collection geometry for ultrasonic
computerized tomography ....................................... 48

vii

Figure 5.1.1 Non-intrusive temperature profile probing system 53
Figure 5.1.2 Typical ultrasound pulse waveform ...................... 54
Figure 5.1.3 Invasive temperature measurement system ............ 57

Figure 5.2.1 The flow chart of scanning and data

acquisition program ................................................. 61
Figure 5.2.2 The flow chart of data averaging program ............. 62
Figure 5.2.3 Program for processing B-scan data ...................... 64
Figure 6.1.1 The set up of experiment 1 ..................................... 68
Figure 6.1.2 A typical echo return with noise ............................ 69
Figure 6.1.3 An echo return after averaging process .................. 70

Figure 6.1.4 The autocorrelation results of the
acoustic velocity change ........................................... 71

Figure 6.2.1 3-dimensional image for flat temperature
gradient in water ....................................................... 75

Figure 6.2.2 Temperature profile of water without the
use of reference signal .............................................. 76

Figure 6.2.3 Temperature profile in water with the use
of reference signal .................................................... 77

Figure 6.2.4 Temperature profile in water by using

fluoroptic thermometer ............................................. 78
Figure 6.3.1 A layered temperature gradient model ................... 81
Figure 6.3.2.a B-scan image to show a boundary shifting .......... 82
Figure 6.3.2.b B-scan image for temperature measurement ........ 83

Figure 6.3.3 Temperature profile of the upper region

viii

of the model .................. 84

Figure 6.3.4 Temperature profiles of the lower region

of the model ............................................................. 85
Figure 6.4.1 The set up for experiment 4 ................................... 91
Figure 6.4.2 The echo waveform of a pork kidney .................... 92

Figure 6.4.3 B-scan image of a pork kidney to show the
boundary shift due to temperature changes .............. 93

Figure 7.1.1 Effect of time-of-flight resolution on
detectability of local ultrasound speed change ......... 97

ix

CHAPTER 1

Introduction

Non-invasive temperature monitoring has many important applications
in areas of hyperthennia treatment of cancer, studying physical properties of

materials, power generation and nuclear reactor safety monitoring.

The use of focused ultrasound to induce localized hypenhermia in
malignant tissue is currently being investigated in many places. The purpose
of these investigations is to selectively damage malignant tissue in order to
inhibit its growth and ultimately cause its disappearance. The ideal tempera-
ture for hyperthermia treatment is to have the tumor at the threshold level
(425°C) or above and the surrounding healthy tissues at normal levels
(37°C ). Temperature distributions have been produced in tissue which come
very near to this desired distribution so that lab experiment may be per-
formed [1].

During hyperthermia the treated region requires accurate temperature
monitoring to ensure that normal tissue is not damaged. Current practice is
to insert thermocouples or thennistors encased in hypodermic needles to
make the necessary measurements. The invasiveness of this technique limits

the number of temperature probes used and in some cases makes it very

2
cumbersome because repeated insertion can result in other hazards such as
infection and tissue destruction. In addition, placing temperature probes into
the treated region often complicates the heating process [2]. A method of
monitoring the temperature rise during hyperthermia without intrusion into

the body would clearly be beneficial.

Non-invasive temperature measurements are also widely used to deter-
mine. properties and states of materials. There has been an increasing use of
ultrasonic temperature measurements for nondestmctive characterization of
material microstructures and mechanical properties. Therefore, it is impor-
tant to have appropriate practical methods for making temperature measure-

ments for a variety of purposes.

A number of approaches to non-invasive temperature measurement
have been used. Among the approaches are those that use single-frequency
continuous waves, tone bursts, and broadband pulse-echo waveforms. In the
latter case, the basic problem is to determine the time delay between succes-
sive echoes, from which the velocity change as a function of temperature
could be measured. The pulse-echo approach using a single transducer as
both transmitter and receiver has many practical advantage over the

transmission methods and therefore will be treated exclusively in this thesis.

The measurement technique developed in this thesis is theoretically

3
based on the ultrasonic velocity as a function of temperature. The technique
is also based on computer digitization of broadband pulse-echo waveforms.
A number of software routines have been implemented in the ultrasonic
imaging system in the Ultrasonic Research Laboratory for the purpose of

non-invasive temperature measurement.

The immediate objectives of the work reported here are to monitor tem-
perature distributions inside a layered material and to obtain temperature
profiles of experimental models with well-defined boundary and temperature
gradients. The work described here provides a base for future research and
applications in medicine and engineering. This research is to seek the feasi-
bility of using ultrasound as a supplemental or primary mode of non-
invasive temperature monitoring during simulated hyperthermia treatment.
Techniques to be developed include the use of higher data sampling rate,

calibration of standard materials for absolute temperature mapping.

The most common used medical imaging techniques use X-rays or
ultrasound. X-ray technology, including CAT scans, and ultrasound tech-
nology present images which essentially show differences in density. This is
generally easy to visualize. ( More specifically, ultrasonic B-scan images
show sharp differences in the acoustic impedance pc , or density times pro-
pagation velocity and radiography using X-ray radiation shows the spatial

distribution of X-ray attenuation coefficients which are highly density

4
dependent.) The difficulty with any thermal imaging technique is to present
an image with various temperature gradients. For infra-red imaging, digital
signal processing techniques are used for color coding to display the regions
of different temperature. This type of imaging display could very well be

employed by ultrasonic non-invasive temperature monitoring.

The contents of this thesis are organized as follows. First, the theoreti-
cal foundations of the temperature dependence of ultrasound propagation are
reviewed. Second, judging from published research results, theoretical con-
siderations for temperature monitoring will be described as the experimen-
tal basis. Third, a variety of non-invasive temperature measurement tech-
niques are briefly described for overviewing and comparison. The remainder
of the thesis will devote to the details of the research work in ultrasonic tem-
perature monitoring, including system configuration, experiment design,
result and analysis. The system limitations and suggestions for future study

will also be discussed.

CHAPTER 2

Theoretical Consideration

This chapter provides the necessary theoretical background for the
proposed techniques of non-invasive measurement of temperature . In the
first section, some related ultrasound principles are reviewed. It is followed
by the theoretical foundations of ultrasonic propagation on temperature
profiles. Some new measurement techniques are proposed and imple-

mented based on literature review and laboratory experiments.

2.1 Review of Ultrasound Principles

It is useful to examine the theoretical foundations of temperature
dependence on ultrasonic propagation. This information is well docu-
mented in the literature and is included here for completeness and con-

venience.

2.1.1 Wave Motion

Ultrasonic energy travels through a medium in the form of shear and
longitudinal waves. The simplest type to be studied is a longitudinal sin-
gle frequency displacement in an isotropic medium which is perfectly elas-

tic. The energy is transmitted continuously, and the vibration is simple

6

harmonic lattice displacement about a mean particle position. A particle is
an element of volume which is continuous with its surroundings, but small
enough for quantities which are variable within the medium to be constant
within the particle. The movement of the particles is resisted by elastic

forces due to the molecular structure of the medium.

Let us consider plane, non-spreading longitudinal waves. The dis-
placement amplitude u from the mean position of particles in simple har-
monic motion, at any instant t, and at a fixed point along the direction of
the wave, where u=0 when t=0, is given by

u = “0 sintot (2.1.1)
where u 0: maximum displacement amplitude, and (0:21: f , where f

is the frequency of the wave.

Velocity is equal to rate of change of position, and so it may be found

by differentiating the particle displacement with respect to time. Thus

v u = uom cosmt (2.1.2)

=3?

Similarly, the acceleration a of the particle towards its mean position
may be found by differentiating the particle velocity with respect to time.

Thus

_9v

a — E = —u omzsinmt (2.1.3)

7

The significance of the negative sign is that the particle is decelerating

as it moves away from its mean position.

From Equations (2.1.1),(2.1.2) and (2.1.3), the phase of the particle

velocity leads that of the particle displacement by a time interval

corresponding to 3%- radians of phase-angle difference, while the accelera-

tion is it out of phase with the displacement.

Ultrasonic energy is transported by mechanical vibrations at frequen-
cies above the upper limit of human audibility. The ultrasound consists of
a propagating periodic disturbance in the elastic medium, causing the parti-
cles of the medium to vibrate about their mean (equilibrium) positions.
The vibratory motion of the particles is essential to energy propagation.
The transmission through the medium is strongly dependent on the
ultrasound frequency, temperature and the state of the medium such as
gas, liquid, or solid. The velocity values listed in Table (2.1.1) are under
the assumption of constant temperature of some fixed value. The table

illustrates the velocity dependency on the medium composition.

Below some temperature, the speed at which ultrasonic vibrations are
transmitted through a medium is inversely proportional to the square root
of the product of the density and the adiabatic compressibility of the

material. The speed of sound ( c ), along with frequency (f) determine

Table 2.1.1 Approximate values of ultrasonic Velocities of various media

[1]

 

 

 

1
Medium Sound Velocity (m/sec)
Dry air (20°C )%343.6
Water (37°C )% 1524

Amniotic fluid 1530

Brain 1525

Fat 1485

Liver 1570

Muscle 1590

Tendon 1750

Skull bone 3360

Uterus 1625

 

 

 

the wavelength (A), from the relationship

7. = 5- (2.1.4)
f
The knowledge of the speed at which ultrasound is transmitted

through a medium is used in the conversion of echo-retum time into

depth of material being imaged.

2.2 The temperature dependence of ultrasonic propagation

It is important to realize that, although energy is transmitted through
the medium as the result of a wave motion, no net movement of the

medium is required for this to occur.

In reality, the whole of the curve in Figure (2.1.1) is moving forward
with a velocity c , which is measured by the distance which any particular
part of the curve travels in unit time. For this reason, it is sometimes
called a travelling or progressive wave. The phase of the wave indicates
in what part of the vibrational cycle the wave happens to be at any partic-
ular instant of time. The transmission of the disturbance is not infinitely '
fast, because a delay occurs between the movements of neighbour parti-
cles. The velocity of propagation is controlled by the density of the
medium and its elasticity. The relationship depends upon the kind of
material and the wave mode. It can be shown that, for longitudinal waves

in fluids

c = — (2.2.1)

where K a = adiabatic bulk modulus, and p = mean density of

medium.

The adiabatic bulk modulus (which is the reciprocal of the adiabatic

compressibility) is not the same as the isothermal bulk modulus K i

10

obtained by static measurements, although for most liquids the two quanti-
ties are almost equal. Thus

K a = 'y K,- (2.2.2)
where y = the ratio of specific heat at constant pressure to that at constant

volume.

In the case of longitudinal bulk waves in isotropic solids, the situation
is complicated by the fact that the shear rigidity of the medium couples
some of the energy of the longitudinal wave into a transverse mode. The
appropriate modulus for calculating the velocity contains not only a term
which depends upon the bulk modulus, but also one which depends upon

the shear modulus. The formula is

c = \l K “:30 (2.2.3)

 

 

where G = shear modulus.

The bulk and shear moduli are related to Young’s modulus Y and

Poisson’s ratio 0 as follows:

_ __Y_
K " \/ 3(1—2o)
_ , / Y
G " 2(1+o)

Substitution of these values in Equation (2.7) gives

6 .-_- \/ “1“” (2.2.4)

 

 

 

 

(1-20)(1+G)P

11

The elastic constants are temperature-dependent, and so the velocity
alters with temperature. The relationship can be quite complicated; for
example, Figure (2.2.1) shows the temperature variation of the velocity of

ultrasound in distilled water at atmospheric pressure.

The temperature dependency of ultrasound has been also investigated
by Isaak Elpiner [3], who provided the basic laws governing the propaga-
tion of ultrasound waves in gases, liquids, and solids.

The velocity of sound waves c in gases is given by the Laplace for-

mula if the amplitude of the vibrations is relatively small:

= 4/15. = «’1.
c p MRT (2.2.5)

~ c
where P is the pressure, p is the density of the gas, 7:};- is the ratio of
V

the specific heat at constant pressure (cp) to the specific heat at constant
volume (CV ), M is the molecular weight, R is the gas constant, and T is
the absolute temperature.

The velocity of sound in liquids depends on the compressibility and

density of the medium and is calculated from the formula

= _L.= _L 226
c Vfiadp Vfizsp (")

where [3 is the compressibility,

[3 = ——- (2.2.7)

12

that is, [3 is the relative change in volume when the pressure changes by
dP; and Bad is the adiabatic compressibility, Bis is the isothermic

compressibility, i.e., the compressibility at constant temperature.

In a slender solid rod the velocity of longitudinal waves is

E
= — 2.208
c ' \l p ( )

where E is Young’s modulus and p is the density of the solid.

In a rod where the cross-sectional dimensions are large in comparison
with the wavelength (or in a solid of large dimensions), the velocity of
sound will not be determined by Young’s modulus, but by the so-called
bulk modulus. For the same material the latter slightly exceeds Young’s

modulus.

The velocity of ultrasound in air is 344 m/sec at room temperature
(20°C).

The velocity of sound in a liquid is generally greater than in a gas
under the same conditions. The velocity of ultrasonic waves in a liquid
diminishes with increase in temperature. Water is an exception. The velo-
city of ultrasound in water increases at first, reaches a maximum at a tem-
perature close to 80°C , and then diminishes with further increase in tem-

perature.

13

The velocity of ultrasonic waves in solids is greater than in gases and
liquids. For instance, the velocity of sound in nickel is 5000 m/sec, and in
iron 5850 m/sec. In heavy. and inelastic metals (lead, for instance) the

velocity of sound is 2169 m/sec.

For soft biological tissues ( muscle, fat, nerves, liver ), the velocity of
ultrasound is 1490-1610 m/sec, while for bone tissue it is 3300-3380
m/sec.

As has been noted by Kinsler and others (1982), "theoretical predic-
tion of the speed of sound for liquids is considerably more difficult than
for (ideal) gases. However, it is possible to show theoretically that [3:157-
where BT is the isothermal bulk modulus. Since no simple theory is avail-
able for predicting these variations, they must be measured experimentally
and the resulting speed of sound expressed as a numerical ( empirical) for-
mula." Based on this point of view, it is essential that empirical relation-

ships or point values based on experimentally determined values be used.

Important results obtained for propagation velocity as a function of

temperature for tissues are presented in Figures (2.2.2 - 2.2.4)[4].

1575

14

 

l550

1525

1500

Volocity.m s"

1475

I450

I425

 

 

 

1400 l l l l
O 20 4O 60 80 100
Temperature °C

 

Figure 2.2.1 The temperature variation of the velocity of ultrasound in
distilled water at atmospheric pressre.

 

SPEED OF SOUND VS. TEMPERATURE
IN VIVO RUNS

 

1600'
.J.‘
o":
0' .-
o ,-
V .’0’ . .'
1590L ,-’
tn ' ’.
\ ’0’
2 . ,°
auscuu . - ' ,-
0 .' . KIDNEY
o I
$1580 .v' ,‘ LIVER _.—._..
Q. 3‘ / SPLEEH --.-.
U3 " BLOOD -.-.- .-
.~/
5'0““0’ ruuon ......

 

 

 

32 36 40 . 44
TEMPERATURE c

(Data from Nasoni, R. L., et a1. 1980)

Figure 2.2.2 Speed of sound vs. temperature in-vivo runs.
(Data from Nasoni, R.L., et a1, 1980)

16

L070!
Paces Uncle
Solon

Sumo! Cord
K0600]

rum

 
 
 

h

a:

0

I
Debno<

 

 

 

 

U
0
H
\
E
x.
I~
‘
U
0
.1
In
5
L500 -
[.460 L ' 1 ' ' J
I6 24 32 40

TEMPERATURE,°C

Figure 2.2.3 Variation of velocity of sound with temperature in various
tissues. (Data from Johnson, S. A., et a1, 1977)

17

O Hose“
0 Parencnymc
A Salt Solution. 2.59/100 a: Son Soouhoa

 

 

 

[£5.80 - O Baczqraund. 0.9g.’tao a: Son Selene.
Q
o L
a
\
E
>~ ..
I. 1.540
'9‘
Q
0
.J
In
> .-
[.500 P L 1 I l J
20 28 35 “

TEMPERATURE,'C

Figure 2.2.4 Variation of velocity of sound with temperature in breast tis-
sues. (Data from Johnson, S. A., et a1, 1977)

CHAPTER 3

Acoustic Consideration For Temperature Monitoring

It is useful to review the literature on the topic of acoustic temperature
monitoring. Conclusions from either theoretical or experimental studies can

be used in the design of our temperature monitoring system.

3.1 Ideal Step-Function Temperature Distribution Model

The ideal temperature for hyperthermia treatment is to have the tumor at
a temperature of 425°C or above and the surrounding healthy tissue at the
normal temperature of 37°C. The temperature distribution for this ideal
situation was assumed using a step-function in some tissue studies. An

example of the step-function distribution is shown in Figure (3.1 .1).[1]

The step-function distribution lends itself readily to calculating tem-
perature changes for non-invasive temperature monitoring because of its
sifirplicity. With the lateral distance of the distribution, 2R , specified by the
rotation of the scanning focused ultrasound, an initial speed of ultrasound, c ,
through the tissue determined experimentally before the hyterthermia treat-
ment or by using an assumed value, the rise in temperature can be deter-
mined from the change in the propagation velocity. The following equation

can be used :

18

19

 

_ 2R _
AC — tl+A¢/md C1 (3.”)

where 2R - outside diameter of the area of scanning focused ultrasound.
t1 - initial propagation time across the distance 2R.
Ad) - measured phase shift in radians.
(ad - angular frequency of the diagnostic beam.
c1 - the initial propagation velocity over the distance 2R 2.

To determine the temperature change by using above equation, the rela-

tionship between the propagation velocity and temperature must be known.

44

42

4O

38

20

 

 

 

DOG LEG IN VIVO
- DEPTH 1cm
i-
4 3 2 I o T 2 3 4

DISTANCE. cm

Figure 3.1.1 The above temperature profile, which closely resembles a
step-function, is the simplest model from which many calculations for
non-invasive temperature monitoring are derived. The most important of
these calculations includes the determination of the required system accu-
racy.

21

3.2 Effect of Realistic Temperature Distribution

If the step-function model for the temperature profile were precise, then
the change in propagation velocity could be used to determine the tempera-
ture rise directly and would involve ultrasonic scanning in one direction
only. Since this is not the case, the next important consideration in the
analysis of monitoring temperature distributions by acoustic means is then
the examination of the temperature distribution as it deviates from a step-

function. In fact, such an ideal temperature distribution is not possible

because the heat flux -k(%-Zl;) would be discontinuous at the boundaries.

That is, the prescribed temperature distribution would be nondifferentiable
at the boundaries and thus could not exist. This is illustrated by noting the

"smooth" edges of the temperature distribution in Figure (3.1.1).

For hyperthermia treatment, the actual temperature profiles are unfor-
tunately not always approximated well by step-functions, due to variable
blood flow through tumors, their irregular shapes, and possible inhomo-
geneities in the associated acoustic and heat transfer properties such as den-
sity, thermal capacitance and conductivity. Thus the average temperature
across a temperature profile may not necessarily correspond to the maximum
temperature as is the case with a step-function-like distribution. Knowledge

of the maximum and average temperatures is important in the treatment of

22
tumors by hyperthermia because of the known variable response of tissues
to thermal doses. This effect is best illustrated by the results of a study done
by Lele (1977) [5] shown in Figure (3.2.1). These results are also used in
the formation of the protocol for treatment of tumors by focused ultrasound.
As a consequence of these facts, it is useful to determine analytically the
average temperature across a more realistic model of the temperature distri-
bution as it relates to the maximum temperature in a step-function tempera-
ture profile. This analytical determination is useful because it indicates the
deviation of the real temperature profile and can subsequently be used to
evaluate the utility of one-dimensional ultrasonic scanning in this applica-

tion.
As the conclusion of the study by Abramowitz, et a1 (1964) [6]. the ratio
of the average temperature in the more realistic temperature profile and the

average temperature across the ideal profile does not differ by much more

than a factor of two. The range of values for TT/Tmam is 0.46 to 1.55. Since
T/Tmax does not differ by much more than a factor of two, the use of the
step-function model to approximate the required system accuracies is
appropriate. In addition, if a thermocouple is used to determine Tmax while a
tumor is being insonated with scanning focused ultrasound, the step-function
model is still a good first approximation. The goal is, however, to use the

change in propagation velocity across the insonation volume as a means to

23
calculate the maximum temperature directly. The conclusion is that if the
temperature profile is indeed nearly a step-function, then the change in pro-
pagation velocity can indeed be used to calculate the maximum temperature
directly given c = f (T). But if the temperature profile evaluated is the more
realistic one, T, the average excess temperature across the distribution, can

only be determined from the change in propagation velocity if:
(1) c =f(T) is known and
(2) c = f (T) is linear over the temperature range investigated.

The value T, if determined, does not give as much information concem-
ing the temperature profile in the realistic model as the value Tmax does in
the step-function model. This fact coupled with the third condition above are

the major obstacles which limit the temperature profile information obtained

by ultrasonic scanning in one-directin only.

24

 

 

 

 

 

 

TEAM-DURATION THRESHOLDS FOR
10 L00 HISTOLOGICAL DAMAGE BY DEEP
LOCALIZED NEATING. uauuaum aura,
In SPINAL CORD. my“. menu. SKELUAL a
g CARDIAC MUSCLE. SKIN, in nu
I

60 L x
.U a F 1‘
1.: ® ' '
a: g I: ' pun“
3 .
"i o
('5 so - '
Q I innate“:
5 cu, ooo. Mount?
.— 2 HUMAN cum, 5. c

INTRA-OPERAIWC
0 ma «or dis!!!“ ®
40 - O PIPPERS or or 'CORNEA
A RODNEY-REC ® ® ®
H ntumours no new: . Sam (9 one orrtcnett came:
0.1 I to: ID Irma. IO the ID
secto“ 10° 10' to2 to’ to‘ to’

INSONATION OR HEATING DURATION

(Data from Lele, 1977)

Figure 3.2.1 Temperature - Duration threshold for histological damage by
deep localized heating ( Mammalian brain, spinal cord, liver, kidney,
skeletal and cardiac muscle, skin).

25
3.3 Effect of the Nonlinear Relationship Between Temperature and

Velocity

Another complication is the fact that many equations relating propaga-
tion velocity to temperature are nonlinear. The influence of this nonlinearity
is best illustrated by examining a simplified model, which exists for the
temperature distributions, for the empirical relationships between propaga-

tion velocity and temperature.

To demonstrate this point, consider a simple nonuniform temperature
distribution as shown in Figure (3.3.1.a). Note that the average excess tem-
perature through the distribution is T. The approximate relationship between
c and T is given in Figure (3.3.1.b). Figure (3.3.l.d) is obtained by switch-
ing the axes of Figure (3.3.1.a) and Figure (3.3.1.b). The estimate of the
average temperature across the distribution is made and is compared to the
average velocity. The simple result of the nonlinearity is that E ref—(T)
although c = f (T) for nonuniform distributions. (The bar notation above
the variables represents the average value of that variable over the distribu-
tion). This is a very important point that does not seem to have been stressed
in previous work done on the determination of c = f (T) in tissues. In fact,
in some studies a great deal of effort was expended using regression analysis
to fit higher order polynomials with coefficients of three significant figures

to the obtained data. It was assumed that the average temperature across a

26
nonuniform temperature profile could be used to determine the propagation
velocity as a function of temperature even if this functional relationship
were nonlinear. This error can further be illustrated analytically. Consider
the functional notation for the temperature distribution and propagation
velocity as a function of temperature given below:

(1) T = g (x) is the temperature distribution.

(2) c = f (T) is the propagation velocity as a function of temperature (non-
linear).
The current average velocity over the temperature distribution from a to

b is:

1 b
75:; if [g(x)ldx (3.3.1)

and this is correct average velocity provided that c = f (T) is known.

If c has a linear relationship to T, i.e.,

C =A1T +A2 (3.3.2)

where A1 and A2 are constant, then the average velocity may also be

obtained from

1 b
dx . 3.3.3
fl'b—_a£g(x) 1 < )
The impact of the nonlinearity between c and T with regard to ultra-

sonic non—invasive temperature monitoring is that when there exists a

27

nonuniform temperature distribution, the change in ultraSonic propagation
velocity cannot be used to determine the average change in temperature

across the temperature profile unless
(l) the temperature distribution is ,in fact known to be uniform or

(2) the velocity distribution is determined by B-scan or C-scan. In either
case, c = f (T) must be known and the distribution prior to insonation

must be uniform and known.

 

 

 

 

7]: __ T(x)
. l ,
X=O x=b x
c 11
b a
T

Figure 3.3.1 (a and b) Illustration of the problem associated with deter—
mining sound velocity as a function of temperature using the average tem-
perature across a non-uniform temperature profile.

 

 

 

 

 

T 4’
6.6c
c
T t A
C(x)
'5' __...
T __ T00-a
d i \
o x b

29

T = 8(4)
c =f(T)
6(4) =11 8(4) 1
1 b
figment: <1)

2:-

b
1
M11 [gram-1 (2)

a

 

(T:

Figure 3.3.1 (0 and CI) Note: This is generally true for nonuniform tem-
perature distributions and nonlinear relationships between c and '1‘. 111 some
lmnted eases (l)=(2), but (1) is the correct expression for Fund (2) is not.

CHAPTER 4

Velocity Measurement Techniques

A variety of techniques have been developed for measuring ultrasound
propagation velocity. Methods are used for either detecting temperature

variation or determing characteristics of materials.

4.1 Review of Temperature Measurement

4.1.1 Interferometric techniques

Interferometric techniques are usually use continuous waves and only
one transducer in a reflection mode. The velocity is'measured by setting up a
standing wave between the transducer and the reflector and varying the dis-
tance between the two, producing minima corresponding to integral numbers
of half wavelengths. This technique relies on an accurate measurement of
the distance between reflector and transducer and also on accurate measure-
ment of the frequency. The velocity is then calculated from the known fre-

quency and measured wavelength.

As noted by Wells (1977)[7], “the accuracy of the method depends on
the parallelism of the opposite ends of the interferometer, and a high degree

of mechanical precision is necessary."

30

31

Interferometric techniques suffer from several drawbacks when
attempting to apply them to non-invasive thermometry. It is difficult to vary
and measure, in living tissue, the change in distance associated with integral
numbers of half wavelengths without altering the physical shape of the tis-
sue. It does not seem to offer the advantages for non-invasive temperature

monitoring during hyperthermia treatment.

4.1.2 Pulse-echo Overlap Technique

The pulse-echo overlap technique has been used widely [8] and has
been analyzed extensively with regard to the effects of diffraction . The
technique relies on measuring the period between a transmitted and received
pulse either in the pulse-echo or pulse-transmit mode. The pulses are typi-
cally overlapped on an oscilloscope and the period measured with a counter.
The period yields the propagation time and with known distance the propa-

gation velocity can thus be determined.

The pulse-echo overlap method can be implemented by analog or digital
instrumentation. In either case, the echoes must have similar waveforms so
that corresponding features can be readily identified and brought into coin-
cidence (i.e., overlapped). The method suffers when pulse-echo signals are
weak or distorted by attenuation or other factors that render them unsuitable

for the overlap approach.

32

4.1.3 The Phase-slope Method

As noted by D.R.Hull (1985)[9], with the phase-slope method, time
between echoes is found by use of phase spectra of echo waveforms. After
the echoes are digitized, a Fourier transform of each is obtained by a
discrete FFT algorithm. The amplitude and continuous phase spectra for a

pair of typical echoes are illustrated in Figure (4.1.1).

After Fourier transformation, both the amplitude and phase spectra are
used to define a central zone within the frequency domain. For example,
this zone may consist of only a narrow range near the center frequency or a
frequency range for which the amplitude exceeds some fraction of the peak
value, and/ or the zone may consist only of the frequency range for which
the phase spectrum is linear. These restrictions eliminate the low and high

frequency extremes where the signal-to-noise ratio is low.

The group velocity is given by

U(f)=7-99—Lz“ 23W (4.1.1)

717'
where f is frequency, 0 is phase angle in radians, and W is the time delay
between echo windows. In—window time delays for each echo can be calcu-
lated by setting W=0 and solving for T(f ) = (d 0/df )/(21t). Assume the

phase spectra are linear functions of frequency in the central zone, their

slopes are constants, M = %7°.-, and the in-window time delays are T1 = 32%

33

M2 ' . .

for echo B1 and T2: -2— for echo BZ. The total time delay rs
11:
T = W+(T2-T1). (Figure 4.1.2)

The frequency domain phase-slope method eliminates problems
encountered in the time domain, e.g., the need to account for echo inver-
sions. In addition, it provides convenient criteria for selecting an appropriate
frequency range in cases where the major portion of the phase spectra of the
echoes are mutually linear. Generally, for nonlinear dispersive cases, the
phase-slope method can determine group velocities as functions of fre-
quency. However, if signal-to-noise ratio is low, as was the case for the

composite samples, poor results are obtained.

\l

TACK

Ulll

34

 

 

 

 

 

 

          

 

 

 

 

1.0 -— 1,0 __
> S t.— > .5
:3 23' L12 ——J
<< 3 :3.
e o .2 e o ___/\ /\~_
-5 l l J -,, 1 1 1 J
' 0 so too 150 200 0 so 100 150 200
TIME. nsec TIME, nsec
(3) Echo 81. lb) (£110 32.
151119;9 60 __
a :3 4° -
>'. 10 .— :5 20 .—
-° :4
3 <
i s— g
5. a
"1 1 °- 'éU ' i | l J
0 20 ~10 60 80 It!) 0 20 40 60 80 too
FREQUENCY. MHz FREQUENCY. MHz
1:) Amplitude spectra. :01 Phase szectra.

Figure 4.1.1 The Phase-slope Method. Typical back-surface echoes, Bl
(solid) and BZ (dotted), and their amplitude and phase spectra.

35

 

 

 

F5 3:1 I :2
e
:5
6
>
me

Time domain trace of princmle ecnoes.

Figure 4.1.2.a Time domain trace of principle echoes. Echo FS is
returned by the front surface of specimen, and B1 and BZ are successive
echoes returned by the back surface. Time T is the round-trip time delay
between echoes B1 and 82.

l—
AAAV
..._ UV

-400__

600-

VOLTAGE . mV
0

 

-600 l l l l
0 50 100 150 200

TIME. nsec

 

Figure 4.1.2.b Result of digital overlap method for determining delay
between echoes 81 (solid) and 82 (dotted) using echoes.

36

4.1.4 Cross-correlation Method

Unlike the overlap or phase-slope methods, the digital cross-correlation
method does not require explicit criteria for accepting or rejecting specific

features in echoes affected by distortion or signal-to-noise ratios.

The method is illustrated in Figure (4.1.3) , which shows the normalized
cross-correlation function for three cases of echo positions. The cross-
correlation function of two time domain signals is obtained by conjugate
multiplication using their Fourier transforms in the frequency domain and
retransformation back to the time domain. The cross-correlation function
possesses a maximum in the time domain. The displacement of this max-
imum relative to a zero reference gives the time interval C, which for the
ideal case should equal T2-Tl as measured by the digital overlap method.
Cross-correlation gives this quantity whether the echoes appear in the same
or separate windows. If the echoes B1 and B2 are separately windowed, then

T==W+C.

For the relatively simple and undistorted echoes of Figure (4.1.3),
cross-correlation gives results similar to those obtained by the two previous
methods. The validity of the cross-correlation method depends on the fact
that the displacement of the maximum of the cross-correlation function

equals the time delay between the two successive echoes.

37

The advantages of the cross-correlation method are apparent when the
signal-to-noise ratio is low and/or random noise is superpositioned on the
echoes. One of the properties of the cross-correlation function is that it is
(statistically) weighted by dominant frequencies common to the waveforms
being correlated. Therefore, it returns a group velocity within the frequency

bandwidth of the signals analyzed.

ECHO BI [CPO l2

H-II-fi 9' \

 

§

 

S

 

 

 

7
%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

am < 4
-m 1
4m - -
0 so to) 150 an 0 so too Tsoxj
1M. nut ll)
1% w a:
caoss-coantwrcu tuwcnm
l n—-C -.1
.s «
12 - ll . 99. 2 nsec
° ~ C - 99. b nsec
I- w . C
'.5 4
-l -
-2m —Lso -tm use 0 so tar no 250
"Mi. nscc
(b) TI <T2.
(two or tree a:
son 1 ,
m i ll —4‘ ‘ I
2 200 1 t— 73—“
3’ °~ “\A‘C'
3 .310 . ,
.m . 1
400 . - #4
0 so too tso 1'1) 0 so too we ll!
lwi. me:
i: W 3.
CROSS°CCRREUHCNFUHC1IGI
1 1—— C—-I
.s «
o _ 12 - Tl - -¢8. 9 nse:
' C - .99. 2 nsec
I- w . C
'.5 i

 

 

 

 

 

.j .3 -
-20) ~50 -too 60 0 50 tm 150 En
"Mi. "sec

(1:) TI>T2.

Figure 4.1.3 Result of generating the normalized cross-correlation func-
tion for two cases of echo positions. The displacement of the absolute
maximum of the cross-correlation function relative to the zero reference
guves the time delay C. For these ideal echoes. C was equal to T2-Tl as
measured by digital overlap within the data-sampling interval error of 0.4

ITS.

38

39

4.1.5 Acoustic phase shift technique

An acoustic phase shift technique for non-invasive measurement of
temperature changes in tissues was developed by B.J.Davis and P. Lele
(1985).[10]

The phase shift of an interrogating pulsed ultrasonic sine wave (diag-
nostic beam) is used to measure changes in its pr0pagation velocity through
the heated region corresponding to the temperature change. The system
operation is in pulse-transmit mode, which use two transducers, for
transmitting and receiving respectively. As shown in Figure (4.1.4), (4.1.5)
and (4.1.6), the data recorder is triggered from the transmitting pulse and has
a variable time delay which is held constant during and after the hyperther-
mia treatment. The change in propagation time is measured by determining
the change in total delay time. The change in propagation velocity Aa and

other known parameters is given by the following:

._ 2R2
AC --t1+—E?-Cl (4.1.2)

where t1 is the initial propagation time, At is the change in propagation
time, Cl is initial propagation velocity over the distance 2R2, 2R2 is the dis-

tance over which the temperature rise occurs (insonation pattern diameter).

40

-2282: sham SEE 8% 53330 829?. E aoocoo 2F

02:23: «GALE.

 

mmoz<I0 wmak<mwa2mk

O... ombfiwm m. m2... 20:.

t<o<aoma z. woz<IO ml... 1. N

E a ,. >.. m
/\.I .

_ 5.

ll

 

  

 

 

 

 

 

 

 

.wEUE
4.2. 2:5

ILSPII

02F<m2 m0"—
DZDOm<mP43

 

      

EmOuw><>>

 

 

 

 

 

 

 

 

 

98:26
pzmazék 35$: 38%
828mm x
mmoaomzqgt\ |\ A. »

ZoermEHmE mmDH<mwd2wk

owmaoou oz_zz<ow

w><§
wzfi ommqad
QwHtEmZ/‘mh

IL<<¢II

.mwoaomzxxm...

41

.mouamuoqEB 208:6

024 Sec £58032: 820:6 o>c .8 Soto “tam 89E 0:4 ouswE

AmHoHE more 8:83.055. E:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7..” 1 i t t Gaunt 1 - - (GUMm 1 CONN. OO.m
q q a d - 1 l4 1 - _ - - A t t t l” B+N4t
.. it; . “a. .2 1..
_ . L
_\Ve_. f. p: C; 5 “8.9m-
:__ r: f .1... __ ....
.. 2. j . _____. i» 2
m: 1 .12 _ _ w._ s; in 56.?
a; m. i w. _:__ _.__.._....,_ -.
fl __ _ _ _M @. _._ _ _.________ a. .\ .n/tuwmfiem 85.0
_ :1 .. ._ r. E
. A E“ 2...; 3 1 ,._ M
.._ Z _ i g _ -U as...
._ i. w __ 2.,
._ .__.___ f” P w ..
” _____ ,1 w _ .U 5......
__ Z. ; Q ...
. c. m
.. .84..
83> Qq

MQLWEHELH

42

Adam: .55 93 282: .«o @5532: c888
magnum £3, 59885 38:835.“ mew—.6 3288 203 mew—£0225
26% 2.5 Saw—383. .23 £5 83m 05.59% 0.34 anfi

03E .
68;: E mazooumtomuE. wt:
84. Ed. -85 BE ave. - am.» I one.

«O+®.mt

~®+D.VI

I
Aml “4’. l‘_ - 2

00.0.0

LIL-2L2

ll-

~O+O.7

A A A

l

dO+O.m

A
L
.L
A
A
L
L

 

 

... 8&4
34% Pa
82355

43

4.1.6 Other Techniques

Greenleaf, Johnson, and co-workers [11] have shown that a reconstruc-
tion method similar to that used by X-ray reconstruction may be used to
reconstruct the spatial variation of attenuation and refractive index in a spe-
cial class of data collection geometries. The spatial distribution of the value
of many measurable parameters may be determined by inverting the line
integrals of a related acoustic variable along a set of refracted and/or
reflected ray paths; in particular, the propagation time along a ray is the line

integral of a specific function of acoustic refractive index.

Based on above theoretical approach, which indicates that high resolu-
tion reconstruction images may be obtained by accounting for the refracted
or curved ray, a joint program between S.A.Johnson at the Mayo Clinic and
DA. Christensen at Utah resulted in experimental evidence that internal
temperature distributions of static materials could be reconstructed with
ultrasound techniques. An example of the reconstruction of the temperature
of three water-filled balloons was demonstrated. The experimental

configuration is shown in Figure (4.1.7).

Ultrasound reconstruction methods for determining temperature depend
on theories developed for analysis of transmission data. These theories limit

the use of the body, such as the breast, where reflection and absoption

44
effects are minimal. This technique also suffered from the inaccessibility of

in-vivo experiments in medical applications.

A method used by Dunn and Fry (1961) to determine the acoustic velo-
city in the lung relied on determining the change in acoustic impedance, pc ,

from the reflection at a lung-water interface.

Barlow and Yazgan (1965) determined the phase difference between a
modulated r.f. pulse propagated through a known liquid path and the
incident pulse reflected from a solid-liquid interface by cancelling each
pulse separately against a continuous-wave signal which was adjustable in
phase and amplitude. From two such measurements at slightly different fre-
quency the total phase shift in the liquid is calculated and then related to the

propagation velocity.

45

905:8 can 538:8 See
.9533 Eafiwofioh wontousanU 925355 «.546 oSwE

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

fill] 9:20: -
- 23m :m...l|lL : 53:. d w
:3qu ...vv .1“ vvv mimumm
eouomsk
.6523 a 53$.
:33... .3. 2.2% 3.3.5
.3» 328 A ,
t -_ J.
2 \ ~n\s r :20: 2:28

 

 

22.:

.28 . .
a _o __ 2...:

3:25 bcu cotuozou 3.5.
”Emsmxm xtmvéuoxak QwN~mmgbmzou Ozaomquq:

<b<QkMkZ~ _

 

 

 

 

 

 

 

 

46

4.2 Temperature measurement technique in this research

The ultimate purpose of the temperature measurement system is for
medical applications such as hyperthermia treatment. Such techniques as the
interferometric technique and cross-correlation technique are often claimed
as having a high degree of accuracy and precision. But these techniques are
not well-suited for this application, because they require a relatively noise-
free acoustic environment for their successful operation. Since transmission
mode measurements suffer from the inaccessibility of in-vivo experiments,
an alternative pulse-echo mode measurement has been chosen in this sys-

tem.

The pulse-echo technique used here is also under the assumption that
the speed of ultrasound as a function of temperature is given by an expres-
sion of the form:

V = Vo+mT+qT2 (4.2.1)
It has been observed that V0 is not constant for all tissues, while m is nearly
constant for all tissues from 37°C to 44°C; which is the temperature range
in hyperthermia treatments. The parameter q is very small in this tempera-
ture range for most tissues thus the velocity is linearly proportional to the

temperature T.

Acoustic velocity measurement techniques generally are able to meas-

47
ure changes in velocity much more accurately than absolute velocities. Con-
sequently, with regard to non-invasive temperature measurement, a change
in temperature is measured much easier than is the absolute temperature.
This fact is particularly important if one considers that the temperature
coefficient of velocity is fairly constant for most tissues, whereas absolute
propagation velocity is known to vary. As a result, information regarding
initial body temperatures prior to therapy is important in order to determine

absolute temperature in tissues.

The temperature change is determined as follows:

Vorig = V0 + m Torig

and

Vnew =Vo +m Tnew

are substracted to give:

Tnew = Torig +‘,}T[Vnew "Vorig ] (422)

where T, M, and V were defined previously and "new" refers to the new
temperature to be measured and "orig" refers to the original known tempera-
ture. The value of "m" in the above equation could be assumed as the aver-
age for the representative tissues. It is concluded that the temperature
change is also linearly proportional to the velocity change which can be
measured by the ultrasound pulse-echo techniques outlined in this system.

The basic arrangement of the system and its operations are discussed in

48
Chapters 5 and 6.

Transmitter Locus

 

Profile Data
0 9’ P ”-‘2’
’ a

     

Profile Data

P (Go 9!) Receiver Loci

Figure 4.1.7.b Data Collection Geometry For Ultrasonic computerized

Tomography. (Geometry of two typical transmitter locations, dark and
light lines. Note separated locus and centers of rotation of transmitter and
receiving transducers).

CHAPTER 5

System Configuration and Operation

5.1 Hardware Description

In this project, both a non-intrusive temperature measurement system
and an intrusive temperature measurement system were built for make com-
parison with each other. We used the intrusive method to verify the non-

intrusive method where it is applicable.

5.1.1 Non-intrusive temperature measurement system

The ultrasonic scanning system used for non-intrusive temperature
measurement is shown in Figure (5.1.1). The transducer is used for both
ultrasound pulse transmitter and receiver. It is a long focused, pie‘zo-electric
transducer with 2.25 MHz resonance frequency. The pulse waveform is

shown in Figure(5.l.2).

The stepping motor and its controller is used for 2-dimensional scan-
ning. The scanning part uses two synchronous stepping motors and the scan-
ning resolution is about 1 mm, and the scan area is about 128 mm. by 128
mm. By using 2-dimensional scanning, a temperature gradient (or distribu-

tion ) can be obtained.

49

50

A panametrics 5050 Pulser is used as a pulse generator and echo
receiver. The Receive/Transmit mode of the pulser is used in the system
operation. For transmit, the pulser portion of the 5050 Pulser is controlled by
the external trigger, which is generated by the RANGE -GATE . After excit-
ing a pulse to the transducer, the 5050 Pulser switches to the receive mode .
The receiver portion of the 5050 Pulser is a 40 dB amplifier, with a three
position attenuation control, which allows the input signal to be attenuated
prior to amplification. The receiver output is limited to 2 Vp_p which meets

the requirement of the A/D converter.

The RANGE-GATE is a time-delay controller that controls the time
delay between the external trigger pulse and the sampling enable signal.
Because of the limited buffer size, which is 1024 bytes, it is useful to receive
all of the echos from the specimen in the time interval of propagation from
the transducer to the specimen. The delay time in sample time intervals can

be anywhere from 1 to 255.

The sampling and A/D conversion board were specially designed for
this temperature probing system. The system core is a TRW 1048 monol-
ithic video A/D converter. It is capable of sampling at rates up to 20 MHz,
which is sufficient to sample a signal up to 5 MHz. Dual clocks are used in

this board, one at 14.667 MHz for sampling and conversion and the other at

51
100 KHz for transferring data from the A/D buffer to the system controller (
Crommemco Z-2D micro-computer ) memory through the 8-100 bus. The
reason of not using 20 MHz clock for the sampling rate is that the system is

more reliable at the signal frequencies of interest which range from 0.5

MHz to 5 MHz.

The system controller is a Crommemco Z-2D micro-computer, in which
programs are written in Z-80 assembly language. One of the testing modes
is a one-shot mode, in which the transducer launches a pulse signal to speci-
men and receives multiple echos back from the specimen. After being digi-
tized, all the echo data are stored into a data file by the computer. The data
obtained from this mode can be used for determining the temperature

coefficient for velocity by signal processing to be described in Chapter 6.

Another mode is the Scanning test, which produces the temperature
profile of a specimen. The algorithm at each physical position is the same as
that of the one-shot mode described above. A scanning motor control pro-

gram lets data at every position within the scanning area be stored .

An IBM-PC computer is used for purposes of post-data processing and
graphic display. An Emulation program has been used for interfacing the
IBM-PC and the Crommemco Z-2D. The Emulation program handles the

communication between the two computers through a RS-232 interface. We

52
can use the IBM-PC either as a terminal of the system controller to issue all
the operation commands to the Crommemco Z-2d or as a data-processor to
evaluate the velocity change which indicates the temperature change. After

obtaining the scanning data file, a display of the results may be performed.

53

.82»? wcfioa 0805 2880068 37.005002 fifim 2&5

I‘ll

M772. MPH/‘3.

003000 . ,
_mmqoou#

HI.
HL

 

a:

 

 

haw—Rm: #

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmuaamz<¢e
0.4.2308.
0530
lllllll J
u a
N 00.58200
0200 02 0 ”0,3505 0
8.0
a 0:2 .1... 023.3% f. ”28.55200 «0390.200
0m
030m ”5.82805
.2205.

 

 

 

 

 

 

 

 

 

 

 

* m

54

Amplitude

 

 

 

 

 

 

 

 

 

 

Display range [0, 19991 Time

Figure 5.1.2 The pulse waveform in time domain used in non-intrusive
temperature measurement system.

55

5.1.2 Intrusive temperature measurement system

As a supplemental part of the project for comparison purpose, an
intrusive temperature measurement system was built by using a fluoroptic
thermometer, a commercial unit named Luxtron Model 750, which is a
state-of-the-art thermometry system. The system is shown in Figure(5.1.3).
This machine uses several optical fiber probes to measure different tempera-
tures at the same time. The probe has temperature-sensitive phosphor
material coated to its tip. In operation, the phosphor material is excited by a
xenon flash. The induced flourescence starts decaying with time. The rate of
decay is temperature dependent. By evaluating the decay rate, temperature
can be measured with high precision (within a fraction of a degree) [15].
The temperature read out can be displayed on the front panel and also can be

transmitted to a computer for post-processing.

Instead of using an ultrasound transducer and the Pulse
transmitter/receiver, the Luxtron model 750 was connected directly to the
system controller- Crommemco Z-2D with RS-232 serial ports. The tem-
perature data read by the probe are transmitted to the system controller and

saved for later processing.

In operation, a probe is moved by. the scanner to make a 2-dimensional

scan and obtain the temperature distribution data within the scanning area.

56

The system controller stores the data and then send it to an IBM-PC for 3-
dimensional display.

A program was developed to make the Luxtron model 750 totally

remote control, including the setting of all operation modes and parameters.

 

 

 

 

 

_. micro-comput-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Stepping - Fluoroptic ther-
motor __ mometry:
scanner Model 750

Optical fiber

probe

Water
Reflector Q
S
Cooler

57

 

Crommemco

CI‘

 

 

 

 

 

PC computer
for 3-dimen-
sional display

 

 

 

Heater

Figure 5.1.3 Block diagram of invasive temperature measurement

system

58

5.2 Software Description

The software used in this system are classified into two groups. It is
convenient, for purposes of description, to consider the ultrasonic tempera-
ture measurement system as two distinct parts, one is scanning and data
aquisition, the other is data processing and imaging.

The main flow charts of the scanning and data aquisition programs are
shown in Figure (5.2.1). Note that each data record has three dimensional
location information, say X, Y and Z. From the data stored on disk it is pos-
sible to extract the X and Y co-ordinates of each measurement point, and the
Z co-ordinate as a boundary location. The Z co-ordinate is actually meas-
ured in terms of the time it takes the ultrasound to reach a boundary in the
specimen and reflect back and, if the distance from the boundary to the
transducer is relatively constant, this is proportional to the sound velocity.
Thus, the comparison of time change will be converted to the velocity
change and then mapped to the temperature change if the relationship of

velocity and temperature is known.

In the group of data processing and imaging software the program for
processing one-shot data, basically doing correlation and averaging as
shown in Figure (5.2.2), is used for getting rid of noise. Since the correlation

calculation can distinguish the time difference between two signals which

59
are offset by one sampling point due to the triggering uncertainty, this noise
can be deleted. The noise free, averaged signal can be then processed
(autocorrelated)to find the time interval which provides an accurate measure

of velocity .

After running the scanning and data aquisition program, the data file
obtained is processed in a PC computer to get a three-dimensional image.
The program can find the time interval between boundaries at each position.
This data file can then be displayed as a temperature profile on a cross-

section of the layered model.

The main data processing program is implemented in a PC computer to
process the data file obtained from the scanning and data aquisition program,
which has three dimensional boundary location information. This program is
ideal for experimental models which have well defined boundaries. The pro-
gram reads the source data file and choose a certain Y position to process the
correspondant cross section for a B-scan image. The program reorganizes
the data into groups on different layers. After user specified layers, the dis-
tances between the layers for every lateral position is calculated, and the
results are stored into a new file, a layer distance file, which can represent
the time interval between boundaries at each position. In applications,
scanning will be done twice prior to and after the heating process which will

produce two data files. The B-scan image of these two files will show the

60

time intervals between two layers for all scan positions and will indicate the
velocity change inside the experiment model. This B-scan image can be
interpreted as a temperature profile cross section. The flow chart for the

software is shown in Figure (5.2.3).

jSTART’

4
INITIAL SYSTEH

 

 

 

 

 

A

 

 

 

PROCESS KEY-IN
PARAMETERS

 

 

 

 

 

 

 

MOVE HOTORS TO
THE DESIRED POSITION

J

SAMPLE ULTRASOUND
DATA

 

 

 

 

 

 

 

 

 

 

RECOGNIZE ECHO AND
DISPLAY BOUNDARY

 

 

 

M7

 

Figure 5.2.1
program.

COHPLETE ?

yes

 

i
PRINT COHPLETB
HESSAGB

 

 

 

 

 

 

 

 

RETURN TO SYSTEM

 

 

 

 

END

The main flow chart for the scanning and data

61

aquisition

 

 

Input parameters;
1. the number of data files to be averaged.
2. correlation range

 

 

 

 

 

Read the first data file into an
array

 

 

 

 

 

Do loop: from n=2 to n=N

Read the nth data file;

Calculate correlation to get the difference
between two signals;

Two signals be averaged into one averaged
signal;

 

 

 

 

 

Store the averaged signal data into file.

 

 

Figure 5.2.2 The flow chart of the data averaging program.

62

63

5.3 The System Operation

When using the non-intrusive ultrasonic temperature measurement sys-
tem, a PC computer is used as a terminal to issue all the operation com-
mands. Initially , an Emulation utility is run to set up the connection to the
system controller, the Crommemco Z-2D, in order to manipulate the meas-

urement operation through the terminal.

After setting the scan area parameters, which are X and Y axis range,
step length, the sampling delay time and the echo signal detection threshold,
the scanner translates the transducer over a specimen in a degassed water
tank. A pulse wave is emmitted at each scan position by the 2.25 MHz trans-
ducer. Echos are received and are digitized to detect the boundaries by the
data aquisition program, and only the boundary position value are stored.
The data are stored in the format of (X,Y,Z). The change of boundary posi-
tion in the pulse transmission and receiving path indicates the change of
velocity. If we can get rid of other factor which could affect the boundary
position, such as system uncertainty and the movement of the specimen, the

boundary position shift can be mapped into a temperature change profile.

After the system controller completes the scanning, the data file is
transferred to the PC for post data processing. The imaging software will

construct an image of the three-dimensional data stored during the scanning

64

process and will project it onto a two-dimensional viewing screen.

Stan

 

 

Do 100p: read data from files.

 

 

 

 

 

Choose a cross section to do B-scan image.

 

 

 

 

 

Regroup the data according to layers.

 

 

 

 

 

List data according to layers or make hard copy .

 

 

 

 

 

Input parameters for choosing two layers.

 

 

 

 

 

Count the distance between the two layers.

 

 

 

 

 

Store data into a file to be displayed as a B-scan image.

 

 

 

 

End

Figure 5.2.3 The program for processing B-scan data to find the veloci-
ty change due to temperature change.

CHAPTER 6

Experiment Design and Procedures

A set of experiments were done to verify the system capabilities of
temperature measurement and to prepare for further biomedical applica-
tions. Experiments were conducted in which the temperatures were varied in
a, medium and the change in ultrasonic propagation velocity across that
medium was recorded. These experiments will be described in the sequence
of experiment configurations, data processing, results and discussion which

follow.

6.1 Experiment 1: The ultrasound velocity change of oil due to tempera-

ture variation

This preliminary experiment is for verifying that the acoustic velocity
changes in liquid due to a temperature variation. As mentioned before, the
velocity of ultrasound in water increases with increasing temperature ini-
tially, reaches a maximum at a temperature around 80°C , then decreases
with further increase in temperature. For most liquids, however, the velocity
of ultrasound in the liquid decreases with increase in temperature. Veget-
able oil was used as an example to verify the ultrasound velocity variation

with temperature in liquid.

65

66
The experimental set up is shown in Figure (6.1.1). A 2.25 MHz
focused transducer is immersed in an oil box at a fixed position. A heater is

used to increase the temperature of oil.

A mask, with a small hole which lets ultrasound through, is mounted on
the transducer to provide a reference signal. The transducer launches a
pulse and receives an echo signal. The system digitizes the echos and saves
the data into a data file. Obviously, the time interval between the reference
signal echo and the echo of the box bottom is determined by the ultrasound

velocity. A typical echo return wave form is shown in Figure (6.1.2).

In the experiment, the oil was heated to temperatures of 10°C, 15°C ,
20°C and etc. After an equilibrium is reached at each temperature, the sys-

tem is triggered for 10 data files which are later processed.

Data processing is used to improve the accuracy of the signals. As
shown in Figure (6.1.2), the signal is mixed with noise. One way to increase
the signal-to-noise ratio is to use averaging of multiple returns. The zero
mean stationary noise is averaged to zero so that the signal-to-noise ratio

will increase by W if N signals are averaged.

When averaging multiple returns, the signal phase differences due to
system uncertainty can cause data degradation. Thus signal phase alignment

is necessary before signal averaging is undertaken. By calculating the

67
correlation between two signals, the phase difference is detected. The posi-
tion of the peak value of the correlation function is the distance between
these two signals. These two signal are then aligned according the distance
and are averaged. At each temperature, 10 signals are averaged to reduce

noise and achieve a clean signal as shown in Figure (6.1.3).

For the signal at a certain temperature, the wave propagation velocity
can be calculated from the time interval between the target distance. The
velocity change is related to the temperature change by the time interval
change. To calculate the time interval from the reference mask and the
bottom of the oil box to a precision of 1 sampling point (66 ns), autocorrela-
tion was used for each averaged signal. Calculation results of autocorrela-

tion are shown in Figure (6.1.4).

 

Pulser;

 

Transmiter
and Receiver

 

 

 

 

 

Vegetable
oil

 

Transducer

 

 

 

 

Range gate; System Con-
troler

AID conver- Crommemco

tor (L80)

 

 

 

 

 

 

 

 

PC computer
as a data
processor
and graphic
terminal

 

 

 

Heater

 

 

 

 

 

Figure 6.1.1 Experiment 1 set up. Test velocity dependency
on temperature for vegetable oil.

68

69

.023
-33 8380082 2 0:0 :0 m0 owfino 3322, 0:300“ 05. ~30 000mm;

  

538 0:00 3290 <

70

.3805 wEwEo>m but“ EBB 0:00 5.. 0.0 oeswi

2 >>

 

 

NV duq 0

.dc

 

§

 

 

 

 

 

 

 

71

donut? 2290088 9 0:0 :0 C0
owcano .9023 0300000 2: C0 3.3.8 00220000032“ 05. 04.0 2:me

«>A—> .0 com 3.322.th

 

 

 

 

 

 

95.833»
28.2 TS
~>r¢mu©~w> .0 com "239.380...
.38: _ 1.8 g >
:Eéfibfi _ ular- c. marl;
wu—Smom comum—QCOQOH=< . .
flak-MD?

 

72

6.2 Experiment 2: Temperature profile in water

The purpose of this experiment with water was to verify that the system
could indeed monitor a change in temperature and also could obtain a tem-

perature profile in a medium by means of transducer scanning.

The experimental configuration is shown in Figure (5.1.1). A 2.25 MHz
piezoelectric focused transducer was mounted to a, scanner which was
driven by stepmotors for 2-dimensional scanning in a degassed water tank.
A plastic ice container for cooling and an electric heater were put at two
opposite sides in the water tank. A plastic plate was located on the scan area
to the system reflections from equal distances between the reflector and
transducer. Thus the system should be able to detect any velocity change
due to temperature change when an ultrasound wave propagates in the
medium. The data from every position can then be used to obtain a tem-
perature profile in the medium. Again, a physical reference ( the mask previ-

ously discussed) is used to eliminate uncertainty in signal triggering.

During scanning, both the specimen and transducer are kept under
water in the degassed water tank. The transducer is held above the specimen,
usually at a height determined by the focal length of the transducer. The
transducer is moved parallel to the surface of the plastic plate by two step-

ping motors which control movements in the X and Y directions.

73

Initially, the area of the specimen to be scanned is selected. The move-
ment units are dependent on the step size of the motors , which is about
1mm/step. The spatial resolution is defined by specifying the intervals at
which measurements will be recorded, the minimum spacing being 1mm.

Once the initial parameters are set, the scanning process can begin.

During the scanning, the system controller triggers the transducer and
receives all the echo signals from the specimen. Only the first echo is
identified and stored into a file, accompanied by the X and Y location. This
data file is then displayed as a three- dimensional image to represent the
temperature profile in the medium. Figure (6.2.1) shows the constant tem-

perature profile response.

To obtain a temperature gradient in water, an ice cooler was put on one
side and an electric heater was tumed on at the other side of the tank. A
scan over this temperature profile give the results shown in Figure (6.2.2)
when no reference signal is used. When the reference is used the results are
as shown in Figure (6.2.3). As mentioned before, because of the triggering
uncertainty, the travel time from the transducer to the reflector plate was
affected by both system triggering uncertainty and velocity change. The
triggering uncertainty is removed by the use of the reference signal, leaving

only a variation due to temperature.

74

To compare the non-invasive measurement, with an invasive tempera-
ture measurement, as described in Chapter 5, the image in Figure (6.2.4)
was obtained. The temperature profile for the invasive measurement looks
smoother than that in Figure (6.2.3). One reason for this difference is that
the probe of the fluoroptic thermometer not affected by the heat stream

inside the water tank.

75

 

 

 

 

 

 

 

 

Figure 6.2.1 Three dimensional imaging for the constant temperature in
water, using a reference signal.

76

.3:me

8:288 .8 3.: 05 :55; 833 Mo 0503 2333809 flab unswE

can no:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

98 ECU

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mucouwfih MO 0m: OS“

8053, 8:25 a8 058 oeambmfiofi

77

.3:
-wa 8:882 mo 8: 05 53, 333 E oan 2333th. m.m.o uSwE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 6.2.4 Temperature profile in water by using fluoroptic thermome-

ter.

78

79

6.3 Experiment 3: Temperature profile probing of a layered model

In this experiment, only the non-invasive measurement system is viable,
the invasive measurement system, though it has higher resolution and more
sensitivity, is impractical. The layered temperature gradient model was built
using two plastic boxes. As shown in Figure (6.3.1), one box was glued to
the top of the other. Two 25 ohm resistors connected to a variable voltage
source were used as heaters. These two heater were placed at opposite sides
to allow different temperature distributions. Scanning of the this model was
initially done without a voltage applied. A program was developed to detect
the boundary shift due to the temperature gradient in each layer. Since the
data include 3-dimensional location information, the X and Y represent a
certain cross section in a B-scan, while the Z is the boundary location. Fig-
ure (6.3.2.a) and (6.3.2.b) show the conventional B-scan displays for dif-
ferent temperature conditions. To present the experiment results as velo-
city profiles, the program processes the original data file and calculates the
time interval between any two boundaries and stores it into a new data file
along with the X and Y co-ordinates which then can be interpreted as the
velocity at each position. The velocity change will be found by comparing
the data files at different locations, and these difference can then be inter-
preted as relative temperature difference. The results for the first step are

shown in Figure(6.3.3.a) and Fig. (6.3.4.a), in which the temperature profile

80

shows an almost constant temperature throughout the entire region.

In the second step, 50 volts was applied to each resistor for five minutes
to create a temperature gradient inside the two boxes. As depicted in Figure
(6.3.3.b) and Figure (6.3.4.b), the heater in the upper region was placed on
the right-hand side, thus the temperature gradient results obtained were as
expected. After heating, the temperature increases toward the right-hand
side where the heater is located. By comparing two images, an exact tem-
perature gradient or a relative temperature change can be identified. This
relative temperature change monitoring alone would be a valuable tool for
many applications. For the lower region, the heater was installed on the
opposite side. Similarly, Figure (6.3.4.b) shows a temperature increment
toward the left-hand side because of the position of the heater. The
significance of Figure (6.3.4) is enhanced by the fact that it is a temperature
profile of a sublayer, which is a more difficult region to monitor by conven-

tional methods.

 

81

 

 

 

water

 

\/

 

 

upper region

 

/

 

 

 

 

 

 

\

Q [0% region
/r l

 

\ /

 

 

\/

heaters

Figure 6.3.1 A layered temperature gradient model.

 

82

At low temperature (' 5° C)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Z
«In . v
“SJ-”Mav- a...-
412 i
424 n - .
m:m—VP
436 _-- WH5-s- n
“3 1
4“
O 16 32 48 64x 80 96 112 1238
2 At room temperature (" 220 C)
m
412 1
MW
424
436 _
3.5—..- a WNW.
W HM
«a
460 O 16 32 48 64x 89 96 112 1218
g“ Companson MM
1 __,——H—l-sv-—t__r—
4.12 n1 w - l
. WNEHMM—
424 r 1 At I'OOm r-
—J_r5\-fi—-~M--- l tem CBature
436 a“ H—H—H t. ( ~ 2 C) r
, g-H 4- WW?-
W HM
448 At low -
460 ‘ terrrggrnge
9—": ( ) 48 54,11 16 32 48 54! so 96 m 123
ptenoilZJ X-nin : O Y : O liniISJ X'Iin : 0 Y : 9 12:59:36. 95-31-1988

Figure 6.3.2.a B-scan image to show a boundary shift due to temperature
change (X and Z coordrnates are distance and wave traveling time respec-

tively).

83

4582338 233352 .8.“ ”was: Swarm n.m.m.o oSmE

 

 

 

 

.\
1‘

222383 :52 3
.295 ~23 xon a 58m

 

.533

...O_

 

 

\
l

533 can 8m 5MB “89 a gum

eaoqrqn-na .«nuoruv—

and wad ea so vo av

o u a

o u 53-x

mm on

 

 

32-3.3 .~—"3“nu

x
¢- ~«u «a as vo av

on, one};

«a mu

 

 

 

 

 

9.2822 x:

can

con

ave

a~n

oo~

c.6283; 3.

saw

aem

ave

omn

uo~

84

 

 

 

tempgrature
( F ) B-scan
I
ILZC) *
95 p
H —
7 O weremv—um- - -
I
45 i 1
W
B 32 64 96 128x 160 192 224 258

( 1/64 inch )
(a) Temperature Profile of Upper Region without Heating

temperature
B-scan
125
100 L _.._.H-'

7 5 "I'll —'a'-'I

50 ‘L

 

W
B 32 54 SS 128 163 192 24 256
x ( 1/64 inch)

(b) Temperature Profile of Upper Region with Heating-

 

 

Figure 6.3.3 Temperature profiles of the upper region of the model.

85

Temperature
B-scan
mm =

 

 

Now-

170 i

120 H

|-.In-.—'_.I.}-.-I-I|-

'70

1:""°- r—T‘H

 

20 ‘ l‘
s}

 

 

8 32 64 96 128x 160 192 24 258
( 1/64 inch )

(a) temperature profile of lower region without heating

Temperature
B-scan

~&-*
L “-

 

 

 

 

 

 

 

175 ‘ i.
L'. n
125 .-'I':I I'II t '
u I I a an: I l
I — .--0.I'I ‘
75 H

I
25 ~ 54
I, :5
'n J: _—‘ r 4. x —=1

8 32 64 SS 128x 168 192 224 256

( 1/64 inch)

Figure 6.3.4 Temperature profiles of the lower region of the model.

86
6.4 Experiment 4: Design Considerations and Procedural Aspects for

Biomedical Tissue Studies

Ultrasonic non-invasive temperature measurement has become increas-
ingly important as a monitoring method in hyperthermia treatment. Ultra-
sonic properties of many tissues have been reported by numerous authors
which provide the experimental basis for measurement techniques. Many
techniques developed in principle, however, still have not been implemented
in medical applications despite the efforts of many investigators. The effect
of temperature on ultrasonic velocity in biomedical tissues has been exam-
ined by Carstensen [13] and Bamber[14] and others. Preliminary findings
indicate that ultrasound velocity in nonfatty tissues increases with tempera-
ture in the range from 5°C to 50°C for ultrasonic frequencies between 1 and
7 MHz. Figure (2.2.2) and (2.2.3) show that the expected temperature

coefficients of velocity range is from 0.5 to 1.6 m /s 0C .

This experiment was designed to simultaneously simulates hyperther-
mia treatment and to monitor the change in propagation velocity as a func-
tion of temperature (which could be used for monitoring the temperature

variation during hyperthermia treatment).

The experimental set up is shown in Figure (6.4.1). A specially

designed degassed water tank was used for both hyterthermia treatment

87
simulation and for non-invasive measurements. A 2.25 MHz focused trans-
ducer which was 19 mm in diameter was fixed to the bottom of water tank to
represent the treaunent transducer. A constant amplitude signal generator,
which generates continuous sine waves from 60 KHz to 100 MHz, and a
power amplifier, were used to drive the transducer. The acoustic power out-
put was calibrated experimentally by the radiation force method. A holder in
the water tank is for mounting a specimen between the treatment transducer
and the diagnostic transducer. Similar to previous experiments, a 2.25 MHz
diagnostic transducer of 0.5 inch in diameter were above the specimen. A
modified ultrasonic imaging system was utilized in this experiment. A state-
of-art data acquisition board with a sampling frequency up to 40 MHz was
installed in a PC micro-computer. This data acquisition board has less

triggering uncertainty than the system used in previous experiments.

As depicted in Figure (6.4.1), the diagnostic transducer was aligned
with the ultrasound treatment transducer on the bottom. A pork kidney was
used as the specimen on a holder between the two transducers. A pulse was
launched into the specimen by the diagnostic transducer before treatment
insonation. The data acquisition board received, sampled the echos and
stored them a file for later processing. The real time echo waveform is

shown in Figure (6.4.2). The time interval between boundaries was used to

88
represent the velocity within the specimen, which can be mapped to tem-
perature difference. The boundary defined here is the maximum peak in a
local area, where the local area is defined as a period of 100 sampling points
or so, depending on the echo waveform pattern. The starting point of a local

area is the first peak whose amplitude is above the threshold.

After the first data file is stored, a ultrasonic insonation was done by the
transducer on the bottom. A 2.25 MHz continuous sine wave was used for
a duration of 15 minutes. The reading on the meter of the power amplifier
was 20 watts which corresponded to 1 watt of acoustic power. A second
data file was generated from data taken after the insonation. The velocity-
temperature transformation algorithm described above was used for process-
ing the two data files to find the velocity change inside the specimen caused
by ultrasound induced heating. The results listed below are the boundary
positions before insonation and the positions after. The positions represent
the relative temperature change inside a pork kidney due to 15 minutes of

ultrasonic insonation.

The data before insonation at water temperature of 210 C resulted in the

following boundaries:

boundary 1: 272 samples (each sample equals 25 ns)

89

boundary 2: 517 samples
boundary 3: 1064 samples

The data after insonation at higher water temperature and higher kidney

temperature gave:
boundary 1: 252 samples
boundary 2: 447 samples

boundary 3: 1031 samples

The time interval inside the specimen is 792 sampling points in the pre-
insonation data and 779 sampling points after insonation. Assuming that
the thickness of the specimen (about 2.5 cm) has not be changed, the velo-
city change through the travel region inside the specimen can be calculated
as:

AV = 2.5*10-2 2.5*1o-2
779* 25* 10-9 792* 25* 10-9

It is readily simple to get relative temperature change from this velocity

= 21m/sec.

change if a calibrated relationship of velocity vs. temperature is available.
Note that the surface boundary and bottom boundary are also shifted
because of the temperature change along the water path. This result shows
that the pulse-echo method used here is capable of detecting the temperature

change inside a biomedical specimen provided that it does have clear boun-

90
daries.

In addition to the above experiment, several one-dimensional scans
were done across the pork kidney. A round flat shape, electric heater was
used below the kidney for heating. Again, the first scan was done before the
heater was turned on to obtain a B-scan image. The second scan was done
after the heater was turned on for 10 minutes. Figure (6.4.3) shows B-scan
images of the kidney specimen for both cases. Since the heat conduction
inside the specimen is slower than that of the water outside the specimen,
all of the boundaries are shifted in the same direction which indicates that
the temperature has risen in water, but the temperature change inside the

specimen is not as obvious.

 

91

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L_.J\

 

 

 

 

 

 

EIN 310 1..
RF Power
Amplifier

 

Scanner Pulser Data aqui-
sition
PC comput-
Transducer er
Specrmen
._.__. Holder

\ Transducer

 

 

 

 

 

Constant Ampli-
tude Signal
Generator

 

 

Figure 6.4.1 Experiment 4 set up for biomedical applications.

92

 

 

 

2nd boundary

 

51"“ 511111!“ 112121):

 

ii
i

 

 

 

 

 

ML
,in. - r i \“M - :mt .. tl.
t 513113.41"ng1 “11"";filifililfi‘tl’hn'aw i’rl‘th-Jilul‘ 'Mh\\\’wwmawv 7*
31121:? I
if?
'f
lst boundary 3rd boundary
reference
. transducer
srgnal
at bottom
Display range 10. 19??) tion Hag 22 15:40:40 1989

( A pulse launched by a 2.25 MHz transducer, the echo was
sampled at 40 MHz. i.e. l sampling point = 25 ns. )

Figure 6.4.2 The echo waveform of a pork kidney specimen.

93

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

298 B-scan , fi

368 - —.—___._._.__ - L ' - I, ..

ms _ ----:-- _;:_:_:g_

784 _ _::"I—:-E—:E::=l __ ,

m " 23:2?" ‘‘‘‘ "

1948 16 32 48 64x 33 95 112 128
The image before heater turned on ( temperature 220 C).

380 B-scan

368 ——————-:-..T-_-_.-——-_——-’-Z--=—:- - .- LI

535 _.- ........ -___ ‘.-- .- - - - - _ ._ -- - ..__.._

794 ..____..-.....:. ::--..:_:EE:;-; ; :.: 3.3.1.: _ _

872 -—-- , - ,-.--~.—'_-E—=::: --- ~-

1qu e rs 3?:3'5—1fi4x 89 96 1122—7128

The image after heated for 15 minutes ( temperature 290 C ).

Figure 6. 4. 3 B- -scan images of a
pork kidnc ' s )ecimen to show th -
dary shift due to temperature changes. ) I e boun

CHAPTER 7

Analysis of Experiment Results

7.1 Analysis and Discussion

The experiments with water demonstrated the ease with which a change
in temperature can be detected by a change in the speed of ultrasound. Even
though the experiments here were in a preliminary stage, without calibration
for real application, it is necessary to investigate the accuracy and resolution

possible and consider the requirements for future medical applications.

7.1.1 Accuracy and Time Resolution

The results described in Chapter 6 are actually an average temperature
variation along the wave travel region. The accuracy is dependent upon the

time resolution and the ultrasound wave travel region.

Based on previous results by BJ. Davis [10] and SA. Johnson, the
requirement for time resolution to determine temperature difference of vari-
ous tissue sample widths is given in Figure (7.1.1)[15]. Extrapolation of the
graph in the Figure to smaller ordinat values shows that with a time resolu-
tion of 1 ns for example, a 0.20C temperature change in a 4 mm diameter

region can be detected in a tissue with a temperature coefficient of 3

94

95
m/secoC ( typical for water and biological tissues). With a time resolution
capability of 1 ns, an 8 mm diameter or larger region is necessary to produce
a 1 as change in time-of-flight when the temperature changes by 0.10C

within that region. The graph in Figure(7.1.l) shows the behavior of

__ 5t . . .
D — <V>W, where D rs the smallest diameter of a regron where a

specific change in acoustic speed, (EV/V), due to heating is just detectable
by a time resolution of St. The average speed of sound, <V>, is taken to be
1500 meters per second. In table (7.1.1), the limitations of the system with

respect to detectable temperature changes are presented.

For those temperature change values listed in the table where the
change in propagation time is less than 25 ns (40 MHz sampling rate), the
current system cannot detect the associated temperature changes. Con-
versely, for changes in propagation time greater than 25 ns, the current sys-
tem can measure the associated temperature change. The heated area in the
kidney experiment 6.4 consisted of an area with 2 cm in diameter and had a
path length of about 3-4 cm. The resolution of temperature change in this

specimen would be about 1.00 C .

The final error in temperature determination in hyperthermia production
is a combination of calibration error and measurement error. A precision of

0.20C to 0.50C might be required for useful systems. Clearly larger tem-

96
perature measurement errors can be tolerated below 42°C while greater
accuracy is required when approaching higher temperature where tissue

damage can occur.

Acoustic velocity measurement techniques generally are able to meas-
ure changes in velocity much more accurately than absolute velocities. Some
previous research concluded that the temperature coefficient of velocity is
fairly constant for most tissues, but absolute propagation velocity is known
to vary. From the point of view of medical application, information regard-
ing initial body temperatures prior to treatment is important in order to
determine the temperature rise from the therapy. Temperature measurements
need to be done twice, before and after the therapy, to control the treatment

procedure.

97

Table 7.1.1 Change in Ultrasonic Wave Propagation Times as a Function of
Temperature Elevation and Path Length ( All values in nanoseconds)

Path Length through the Region of Temperature Elevation

 

 

AT(0C) 1.0 cm 2.0 cm 3.0 cm 4.0 cm 5.0 cm 10.0 cm
0.1 0.67 1.33 2.00 2.67 3.33 6.66
0.2 1.33 2.67 4.00 5.33 6.66 13.33
0.5 3.33 6.66 10.00 13.30 16.70 33.30
1.0 6.66 13.30 20.00 26.60 33.30 66.60
2.0 13.30 26.60 39.90 53.20 66.50 133.10
5.0 33.10 66.30 99.50 132.70 165.80 331.70
10.0 66.00 132.00 198.00 264.00 330.00 660.00

53 a 0 r f-
1- > C
,3 § 0.0: -
3. 3 5
o a O.) -
E 3 a a: t- 1
g g 0.0! :- -:
w 3 C '
a a ,. ..
s. S 0003:- -‘
E g 0.00: n
4 2
a 1:: 0.002 - -
1 0.00, r L Llurr r 1 1 ins“
0.0: 0.02 0.05 0.: r 2 s

Figure 7.1.1

 

 

 

 

Otauutrtn cl armor: mm curve: in sou~o spun

Graph of the smallest specific change in acouStic speed 5v/v which
can be detected in a region of diameter D using a time of flight detector
of resolution 5!. The average speed of ulu'asound is taken to be 1500

m/sec.

ultrasound speed change in small region of given diameter.

Effect of time-of-flight resolution on detectability of local

98

7.1.2 Medical Applications

Medical application of the technique is not expected to be as straightfor-

ward as the experiments on biomedical tissues described above due to

p—s
0

temperature gradients occurring within the body,
2. unknown values of the temperature coefficient of velocity,
3. patient movement, and

4. complex boundary identification due to the degree of impedance

mismatch and multiple echo returns.

For the pulse-echo measurement, in order to make relative velocity

measurement it is necessary that:

(1) there exist a strong acoustic impedance mismatch at the treatment
region from which a signal can be reflected,

(2) the capability of identifying boundary position despite complex
waveforms from different tissues (due to the degree of impedance

mismatch, wave dispersion and attenuation).

(3) an initial propagation velocity must be obtained prior to treatment, so
the change in propagation time of the reflected signal can be related to a
change in propagation velocity.

Impedance mismatch beyond the treatment region is assured by the

presence of bone or air. In general, tumors which are accessible for

99
ultrasonic heating or visualization are also accessible for velocity measure-
ment. Some less impedance mismatch problem did happen in experiment 6.4
when a pork liver specimen was used. There was no clear surface boundary
appeared which was necessary for boundary shift detecting. In this case, as a
compensative way, a reference signal might be used to obtain major boun-
dary shifting information.

As has been stated, common values of the temperature coefficient of
velocity range from 0.8 to 1.6 m/(secoC) depending on tissue. In medical
applications, the tumor temperature would be determined non-invasively by
measuring the change in ultrasonic velocity in the tumor, calculating the
corresponding temperature change from the known temperature coefficient
of velocity and adding this value to the known initial temperature assumed
to be at 37°C for most circumstances. For the technique to be exclusively
non-invasive and accurate, a better knowledge of the temperature coefficient

of velocity is necessary.

Another problem which can limit the medical application of the above
techniques is patient movement during hyperthermia treatment. The boun-
dary echo shift is assumed to occur over a constant path length and is con-
sidered to be due to temperature changes alone. If the path length varies,
however, the boundary shift can be misinterpreted as due to temperature

change even if no change has occurred. Even movements as small as a few

100
tenths of a millimeter can lead to false temperature change calculations. A
possible way to overcome this problem is to make velocity measurements
synchronous with the respiratory or cardiac cycles. Another way to limit the
false temperature reading is to ensure that the transducer and the internal
body interface, serving as a reflector for the ultrasonic wave such as between
bone and tissue are firmly fixed with respect to one another. This would
allow some movement of intervening tissue but ensure that the ultrasonic
propagation distance is constant as required. The greater the movement of
the tumor during hyperthermia, the less likely these proposed correction
methods will be effective. Treatment areas which are least effected by the

movement problem are the limbs, lower abdomen, head and neck.

It is possible that in .the cases where knowledge of the temperature
coefficient of velocity is good and patient movement is minimal, that the
described non-invasive technique could be used as the exclusive means of
non-invasive temperature measurement. In those cases where the tempera-
ture dependence is not well-known, the described technique could aid in the
determination of tumor temperature using supplemental information

obtained from invasive probes,

101

7.2 Conclusion and Future Study

Monitoring temperature variation inside material during processing is
quite important in many applications. Since the non-accessibility in those
cases where invasive measurement techniques using thermistors and/or
thermal-couples are impractical, the non-invasive ultrasonic velocity method
is a promising technique. This thesis outlined technique to measure ultra-
sonic velocity changes during hyperthermia (produced by scanning focused
ultrasound). The pulse-echo mode is used to monitor the boundary shift due
to velocity change, indicating the relative temperature change. This tech-
nique will not disturb the environment and will be benefit for patients in
medical treatment. The pulse-echo method will provide greater potential

than the transmitting method since it is more accessible in the human body.

The precision of the temperature measurement is directly related to the
time resolution of the echo. Additional improvement can be achieved by
using the data acquisition equipment with higher sampling rate and less
triggering uncertainty. More experiments with biomedical tissues in-vitro
under simulated hyperthermia, to get experimental models and data for

future in-vivo experiments, are necessary.

To calibrate the measurement system, high precision fluoroptic ther-

mometry should be used for comparison of temperature images. The tem-

102
perature coefficient of velocity in different tissues and the absolute tempera-

ture mapping algorithm must still be refined.

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

103
Bibliography
Lele, P. F., Physical Aspects and Clinical Studies with Ultrasonic

Hyperthermia, F.K. Storm Editor, G.K. Hall & Co. Boston, MA,
1983.

Cetas, T. C. and Connor, W. G., Thermometry Considerations in
Localized Hyperthermia, Medical Physics, 5, p79-9l , 1978.

Isaak E. Elpiner, Physical, Chemical, and Biological Ejfects, Con-
sultants Bureau, N.Y., 1964.

Wells P. N. T., Physical Principles of Ultrasonic Diagnosis,
Academic Press. N. Y. p23, 1969.

Lele, P.P., Thresholds and Mechanisms of Ultrasonic Damage to
Organized Animal, Tissues, In Proceedings of a symposium of
Biological Effects and Characterizations of Ultrasound Sources,
1977.

Abramowitz, M. and Stegun, I.A., Handbook of Mathematical
Functions, National Bureau of Standards, Applied Mathematics
Series, p55, June, 1964.

Wells, P.N.T., Bioinedical Ultrasonics, Academic Press, N.Y.
(1977). '

Papadakis, E.P., Ultrasonic Velocity and Attenuation: M easure-
ment Methods with Scientific and Industrial Applications, Physical
Acoustics Principles and Methods volume XII, W.P. Mason, Edi-
tor, Academic Press, New York (1977).

Hull, D.R., et a1, Measurement of Ultrasonic Velocity Using
Phase-Slope and Cross-Correlation Methods, Materials Evaluation,
v.43, p1455, October, 1985.

104

[10] Brian, J. D. and Padmakar P. Lele, An Acoustic Phase Shift T ech-
nique for The Non-invasive Measurement of Temperature Changes

in Tissues, 1985 IEEE Ultrasonics Symposium, p921.

[11]Steven, A. J., et a1, Nonintrusive Acoustic Temperature Tomogra-
phy for Measurement of Microwave and Ultrasound-induced
Hyperthermia, Journal of Bioengineering, Vol. 1, pp 555-570,
1977, Pergamon Press, Inc.

[12] Carstensen, EL. and HP. Schwan, Acoustic Properties of Hemo-
globin Solutions, J. Acoust. Soc. Amer., vol.31, pp.305-311, 1959.

[13]Bamber, J.C. and C. R. Hill, Ultrasonic Attenuation and Propaga-
tion Speed in Mammalian Tissues as a Function of Temperature,
Ultrasound Med. Biol., vol.5, pp. 149-157, 1979.

[14] Wang, Nai Hsien, Remote Sensing by Acoustic Video Pulse Tech-
niques, A 'Ihesis for Master of Science, Department of Electrical
Engineering and System Science, Michigan State University, 1988.

[15] Johnson, S. A., and et a1, Non-intrusive Measurement of
Microwave and Ultrasound-induced Hyperthermia by Acoustic
Temperature Tomography, IEEE Ultrasonics Symposium Proceed-
ings, pp. 977, 1977.

"irritant