azu_td_1347581_sip1_... - The University of Arizona Campus

INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly fi-om the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may
be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly
to order.
University Microfilms International
A Bail & Howell Information Company
300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA
313/761-4700 800/521-0600
Order Number 1347581
A patient/therapist interactive feedback control system to
regulate temperature and control paia in scanned focused
ultrasound hyperthermia
Nathanson, Scott Mitchell, M.S.
The University of Arizona, 1992
UMI
SOON.ZeebRd.
Ann Aibor, MI 48106
A Patient / Therapist Interactive Feedback
Control System to Regulate Temperature and
Control Pain in Scanned Focused Ultrasound Hyperthermia
by
Scott Mitchell Nathanson
A Masters Thesis Submitted to the Faculty of the
DEPARTMENT OF AEROSPACE AND MECHANICAL ENGINEERING
In partial Fulfillment of the Requirements
For the Degree of
MASTERS OF SCIENCE
In the Graduate CoUege
THE UNIVERSITY OF ARIZONA
1992
2
STATEMENT BY AUTHOR
This thesis has been submitted in partial fulfillment of requirements for an
advanced degree at The University of Arizona and is deposited in the University Library
to be made available to borrowers under rules of the Library.
Brief quotations from this thesis are allowable without special permission,
provided that accurate acknowledgment of source is made. Requests for permission for
extended quotation from or reproduction of the manuscript in whole or in part may be
granted by the head of the major department or the Dean of the Graduate College when
in his or her judgment the proposed use of the material is in the interests of scholarship.
In all other instances, however, permission must be obtained from the author.
SIGNED:
APPROVAL BY THESIS DIRECTOR
This thesis has been approved on the date shown below:
R. B. Roemer
Professor of Mechanical Engineering
Date
3
DEDICATION
This thesis is dedicated to bureaucracy and all the people who made this task
almost impossible.
4
ACKNOWLEDGMENT
I would like to thank everyone who made this work possible. First, Dr. Robert
Roemer for his initial encouragement and then his continual belief that I was capable of
this task. Dr. Kullervo Hynynen for his many ideas, valuable suggestions, and the key
phrase "It is easy". Dr. Hal Tharp for teaching me how to control the world (feedback
not dictatorship). Dermis Anhalt for his much needed technical assistance and his
reminders to everyone that "It just doesn't matter". Mark Buchanan for his practical
engineering knowledge and for his ability to suggest ideas that added many pages to this
document. I only hope that one day I am able to pay him back. Andy Dutton for
friendship and the continual reminders of what life is like with stress. Eugene Gross, who
with a simple suggestion will solve all of your technical problems. Without Eugene, no
one would graduate. It is hoped that one day soon we may all call him Doctor. Matt
Rademacher who has the ability to take life's ordinary events and make them into "Matt
adventures". His adventures are worthy of a major motion picture that would make
Indiana Jones seem like a geriatric documentary. David Lamden for his life long
friendship and expert proofr^eading abilities. And The Big A for a never ending supply
of onion rings that would make Exxon cry. But most of all I would like to thank my
parents who not only gave me the strength to press on but made me possible.
5
TABLE OF CONTENTS
LIST OF ILLUSTRATIONS
9
LIST OF TABLES
11
ABSTRACT
12
CHAPTER 1.0; INTRODUCTION
13
1.1 HYPERTHERML\ AND SCANNED FOCUSED ULTRASOUND .
13
1.2 THE NEED FOR IMPROVED TEMPERATURE CONTROL ....
14
1.3 THE NEED FOR PAIN CONTROL IN SFUS TREATMENTS ...
15
1.4 UNDERSTANDING PAIN
17
1.4.1 WHY WE FEEL PAIN
18
1.4.2 HOW PAIN IS SENSED
18
1.4.3 PAIN, THERMAL STIMULATION, AND DRUGS ....
20
CHAPTER 2.0: SYSTEM DESCRIPTION
24
2.1 EXISTING SYSTEM BEFORE MODIFICATION
24
2.2 LIMITATIONS IN THE EXISTING SYSTEM
27
2.3 NEW SYSTEM DISPLAY
27
2.4 COMPUTER CONTROL OF TEMPERATURE
31
2.4.1 MANUAL POWER CONTROL
31
2.4.2 TEMPERATURE FEEDBACK SYSTEM
33
6
TABLE OF CONTENTS (continued)
2.4.3 SOFTWARE STRUCTURE
38
2.4.4 PROGRAM LOOP
39
2.4.5 MAIN PROGRAM
40
2.4.6 ERROR TRAP
43
2.4.7 HARDWARE FOR COMPUTER CONTROL OF
TEMPERATURE
44
2.5 INTERACTIVE PAIN BUTTON DESCRIPTION
48
2.5.1 PAIN BUTTON ALGORITHM
49
2.5.2 PAIN BUTTON OPERATION
63
2.5.3 PAIN BUTTON HARDWARE
64
CHAPTER 3.0: EXPERIMENTS
65
3.1 PROGRAM VERinCATION
65
3.2 EXPERIMENTS
67
3.2.1 DOG EXPERIMENTS
67
3.2.2 PATIENT TREATMENT
68
3.3 TEMPERATURE CONTROLLER
68
3.4 TEMPERATURE CONTROLLER RESULTS
69
3.5 PAIN BUTTON
76
3.6 PAIN BUTTON RESULTS
78
CHAPTER 4.0: DISCUSSION
82
7
TABLE OF CONTENTS (continued)
4.1 TEMPERATURE CONTROLLER
82
4.2 PAIN BUTTON
83
CHAPTER 5.0: FUTURE WORK
85
APPENDIX A.O: PROGRAM USERS MANUAL
87
A. 1 APPLE COMPUTER PROGRAMS
87
A.2 HP TEMPERATURE COMPUTER PROGRAMS
88
A.3.0 PROGRAM MENU
89
A.3.1 QUIT
89
A.3.2 WINDOW DIAMETER
90
A.3.3 CONTROLLER TARGETS
90
A.3.4ZOOM
91
A.3.5 CONTROLLER IS OFF
91
A.3.6 REDRAW SCREEN
93
A.3.7 RESET WINDOWS
93
A.3.8 MOVE CENTER
93
A.3.9 ALIGN TUMOR TRACE
94
A.3.10 ENTER TUMOR TRACE AND TC'S
94
A.3.11 SCREEN TARGET 42.5 C
98
8
TABLE OF CONTENTS (continued)
A.3.12 T/C VALIDATION
A.3.13 ENTER EDGE POINT
99
100
APPENDDC B; ORIGINAL PAIN ALGORITHM
101
APPENDUC C: WIRING CONNECTIONS
106
APPENDIX D.O: GRAPHICS TABLET AND BOX CALIBRATION
109
D.l ACCURACY OF THE TABLET
109
D.2 DUTY CYCLE BOX CALIBRATION
110
REFERENCES
112
9
LIST OF ILLUSTRATIONS
Figure 1.
SFUS System
24
Figure 2.
Original System Block Diagram
26
Figure 3.
Treatment Display Screen
28
Figure 4.
Controller Block Diagram
34
Figure 5.
Program Loop
39
Figure 6.
Main Program Block Diagram
41
Figure 7.
Error Handler Block Diagram
43
Figure 8.
Modified System Block Diagram
45
Figure 9.
The Anatomy of the Trigger Signal
46
Figure 10. The Anatomy of the Dutycycle Output
47
Figure 11. Window and Sector Power
50
Figure 12. First Window
51
Figure 13. Second Window
52
Figure 14. Possible Pain Location
53
Figure 15. Third Window Location
54
Figure 16. Second Window with Pre-existing Window
55
Figure 17. Overlapping Windows
56
Figure 18. Third Window with Pre-existing Window
56
Figure 19. Press in Existing Window - Part 1
57
10
Figure 20. Press in Existing Window - Part 2
58
Figure 21. Pain Button Algorithm - Part 1
60
Figure 22. Pain Button Algorithm - Part 2
61
Figure 23. Pain Button Algorithm - Part 3
62
Figure 24. Successful use of PI Controller
70
Figure 25. Successful use of the PID Controller (With Dip)
71
Figure 26. Dipless Transition to PID Control
72
Figure 27. Improvement in Variance from the use of PID Control
73
Figure 28. Temperature Histogram of all Controlled and Non-Controlled
Thermocouples in Tumor
74
Figure 29. Temperature Input as Seen by the Controller using ARMA Plus
Max Temperature
75
Figure 30. Breakdown of Recorded Button Presses
79
Figure 31. Patient Treatment with Old Pain Button
80
Figure B1. Old Algorithm for Pain Button
102
Figure B2. First Button Press
103
Figure B3. Relocated Window
104
Figure B4. Enlarged Window
105
Figure CI. Schematic of Dutycycle Controller Box
108
Figure Dl. Test Pattern
109
11
LIST OF TABLES
Table 1.
System/Program Checks for Scan Parameter Identification
65
Table 2.
Display Checks from Dog Trials
66
Table 3.
Thermocouple Information from Dog Trials
66
Table 4.
Power Checks from Dog Trials
67
Table 2.
Pain Button Checks
77
Tabled. Connections Between the Gantry and Vectra
106
Table C2. Connections Between the Vectra and Duty Cycle Box
106
Table C3. Connections Between the Vectra and Apple
107
Table Dl. Graphics Tablet Test Results
109
12
ABSTRACT
An interactive temperature and patient pain feedback controller with manual power
control as a function of scan position has been developed and clinically implemented in
treating cancerous tumors utilizing the University of Arizona Cancer Center's Scanned
Focused Ultrasound Hyperthermia system. The temperature controller implemented uses
a Bang-Bang PID controller in 12 adjacent regions. The controller is shown to be able
to reduce unwanted hot spots without substantially increasing the size of cold spots in the
treatment volume. The addition of patient feedback during treatments has reduced the
frequency of undesirable treatment modifications such as moving the scan or lowering of
ultrasound power. Patients were able to repeatedly indicate the areas of pain with a pain
button during treatments allowing a computer to determine the maximum tolerable power
without excessive discomfort.
13
CHAPTER 1.0: INTRODUCTION
1.1 HYPERTHERMIA AND SCANNED FOCUSED ULTRASOUND
Modem use of hyperthermia in the treatment of cancer began in the late 1800's
when Coley noted that the introduction of bacterial toxins had a therapeutic effect in the
treatment of inoperable round cell sarcomas of the neck (Miller et al. 1977). It was not
known at the time whether the toxins or the resulting high fevers were the source of the
improvements in the cancers. Fortunately for the hyperthermia field, it was soon realized
that it was the fever that was the source of the beneficial results. Fever therapy was then
used for a wide variety of ailments including infectious diseases and even arthritis (MiUer
et al. 1977). Since that time, many different modalities have been utilized for inducing
hyperthermia. Methods that heat large areas of the body include hot water emersion
(Overgaard 1980), convective ovens, heating from magnetic induction by the use of
helical coils (Oleson 1984) and microwave antenna arrays (Herman et al. 1988).
Techniques for site specific temperature elevation include the implantation of
ferromagnetic seeds (Stauffer et al. 1984), interstitial microwaves (Trembly et al. 1896),
radio frequency current (Dewhirst et al. 1982), and external ultrasound (Marmor et al.
1979, Lele 1983) and intracavitary ultrasoimd (Diederich and Hynynen 1989). The use
of ultrasound as a means to treat cancer began over 50 years ago with research falling
into one of three categories; ultrasound alone, ultrasound in combination with
radiotherapy, and ultrasound in combination with chemotherapy.
In a survey by
14
Overgaard 1989, hyperthermia was found to be most effective when combined with
radiotherapy. In the opinion of the author, the optimal technique is dependent on the size
and location of the site to be treated. Because of the wide variety of sites and locations
that are seen in patients, the flexibility of ultrasound has shown great promise in recent
years (Harari et al. 1991, Hynynen et al. 1987, Lele 1975, Shimm et al. 1988).
Ultrasound's flexibility stems from its ability to provide variable power deposition
patterns through techniques such as focusing and/or scanning (Hynynen and Lulu, 1990)
allowing for sufficient power deposition to induce therapeutic temperatures in tissue.
However, the flexibility of ultrasound also makes it difficult to use. Currently, one of
the limiting factors on the further use of ultrasound is the present state of the technology
being used. One of the areas that needs improvement is temperature feedback control of
the ultrasound power because the existing controllers do not adequately deal with the
various conditions and disturbances that are present in the human body. Another limiting
factor in hyperthermia treatments that needs attention is the ability to eliminate and/or
reduce patient pain. This work was undertaken in order to improve the operation of
Scanned Focused Ultrasound (SFUS) in both of these areas.
1.2 THE NEED FOR IMPROVED TEMPERATURE CONTROL
The theoretical goal of a hyperthermia treatment (Roemer, 1991) is to elevate the
tumorous tissue region to a target temperature while leaving the surrounding normal tissue
temperature unchanged. With this goal in mind, a measure of the success of a clinical
15
SFUS hyperthermia treatment can be made by looking at what percentage of the
temperature probes in the target volume are above a therapeutic target temperature e.g.
42.5 °C.
Also, it has been established that there is a link between the minimum
measured temperature and the effectiveness of a treatment when only a few sensors are
monitored (Dewhirst and Sim 1984). Additionally, high temperatures can be destructive
to normal healthy tissue. It is therefore desirable to avoid high temperatures in normal
tissue and avoid low temperatures in the tumor while also attaining a uniform temperature
field in the tumor (Roemer 1991). However, as noted by Marmor et al. 1979. achieving
uniform temperatures can be difficult.
In order to improve the temperature field,
feedback controllers have been proposed for use in hyperthermia (Babbs et al. 1986,
Knudsen and Heinzl 1986, Knudsen and Overgaard 1986, Nikawa et al. 1986) and
several for use in SFUS (Johnson et al. 1990, Lin 1990, Lin et al. 1990, Roemer et al.
1986).
1.3 THE NEED FOR PAIN CONTROL IN SFUS TREATMENTS
Dining a SFUS treatment a patient can experience sharp pain when the ultrasound
strikes a bone or a nerve. Bone pain is caused by excessively hot temperatures due to the
high ultrasound absorption coefficient of bone (Hynynen and DeYoung 1988). Nerve
pain is due to the temperature rise of the nerve caused by the surrounding elevated
temperature field or by direct ultrasonic heating of the nerve (Lele 1963). In past
treatments with SFUS, when sharp pain occurred the operator had to rely on the patient's
16
verbal description of the pain's location. This can be inaccurate due to the patient's
difficulty in locational pain judgment. In response to the patient's pain, the operator was
required to lower the overall ultrasound power to alleviate the pain or to move the scan
away from the area. (The same is true for other methods of heating.) However, this is
counterproductive to the therapeutic goal of the treatment - to reach and maintain a target
temperature throughout the treatment volume. Moving the scan or lowering the overall
power can also cause a portion of the tumor to be missed or under-heated.
Since pain can be caused by the ultrasound striking bones or nerves, it would be
best to avoid doing so. In order to selectively heat the tumorous areas, a comprehensive
model of the site to be treated is required. However, such a complete picture of the
treatment site and the area surrounding it are not always possible regardless of the
imaging technique that is used. Even if a complete picture can be compiled, the current
state of technology does not always allow for optimal power deposition (Roemer 1991)
to avoid inflicting pain. Given that most hyperthermia treatments will treat to a patient's
tolerance for pain (regardless of the method of heating) it is not surprising to leam from
a successful SFUS clinical trial, that in 169 of 225 (75%) treatments smdied, the
temperatures achieved were limited by a patient's tolerance to pain due to excessive
temperatures (Harari et al. 1991). This limitation has also appeared in many other
studies. Connor et al. 1977 stated that ultrasound near bone is a major problem that
needs addressing. In another study (Corry et al. 1982) it is noted that "... the most
significant and limiting toxicity (to continuing a treatment) was pain associated with
..(bone)". That smdy also recognized that a major barrier to the advancement of
17
ultrasound for hyperthermia is patient pain. Out of the 31 patients that entered that study,
40% experienced pain with half of those experiencing "treatment limiting pain". All of
the patients whose treatments were limited by pain were treated near sites containing bone
or nerves. This limitation often manifests itself in one or two local areas of the patient
requiring the overall power to be lowered throughout the treatment volume with a system
that lacks flexible power deposition. Conversely, a cool spot could not always be given
more power without causing excessive heating in a different area of the scan. Excessive
heating has been pointed out to lead to "intolerable pain" in the clinical trials conducted
by Corry et al. 1982.
These limitations detract from both the uniformity of the
temperatures and the percentage of temperature probes above the target. In addition, the
operator was required to continually monitor temperatures and adjust the power delivered
to respond to a variety of disturbances, including physiological changes and patient pain,
in order to try and achieve the best temperature profile possible. Therefore, patient
feedback could improve the patient's comfort and the effectiveness of the treatment.
1.4 UNDERSTANDING PAIN
The goal of a hyperthermia treatment is to achieve the best possible heating of the
target volume without inflicting intolerable pain to the patient. Since the operator of a
SFUS treatment often treats to the patient's tolerance for pain, this presents the operator
with several problems which include: how to identify the pain, the determination of
18
important factors in pain sensation, and when does pain become excessive? In order to
answer these questions it is necessary to look at the mechanisms behind pain.
1.4.1 WHY WE FEEL PAIN
The word pain has its origins in the Greek word for penalty (Webster 1981) which
indicates that pain is an miwanted consequence of ones actions. This leads to one of the
main reasons that pain exists in namre. to alert the body of impending damage. The
sensation of pain can be traced back in evolution over the last 250 million years where
the first modem pain receptors have been found in mammals of that era (Miller 1967).
The reasons behind evolution's drive to develop such complex mechanisms as pain
sensations can be seen if one realizes that the sensation of pain is the motivational force
to drive organisms into action aimed at stopping the pain as soon as possible (Miller
1967) and thus prolong survival. A detailed accoxmt of the biological mechanisms of pain
in man can be found in Kiff 1974 and many other sources, therefore pain sensation will
only be briefly summarized here.
1.4.2 HOW PAIN IS SENSED
Pain sensation is vastly different between skin stimulation and deep structure
stimulation. The skin is the body's first line of defense against damage and thus has the
ability to accurately localize and categorize pain.
An animal would develop accurate
19
skin sensation because it has provided a survival advantage to the species in the past
(Moran 1950) by alerting the animal before the source of the pain inflicts a debilitating
or mortal wound.
This is in contrast to pain felt deep within the body which is
commonly sensed away from the source of the pain (Moran 1950, Iggo 1972). For
example, a heart attack victim commonly feels pain in the arm instead of the heart.
While the phenomenon of pain referral is not completely understood it is speculated that
it is a case of "faulty localization" by the brain (Moran 1950). Accurate pain sensation
deep within the body has never evolved because there has never been an evolutionary
reason to do so (Moran 1950). The only deep pain information that is useful to the
survival of an organism is the existence of the pain.
When a painful stimulus is applied to the body, a threshold level must first be
exceeded which causes the firing of nerve impulses. Once activated, it is then possible
to see a double response to the pain (Livingston 1943). The first response is the "fast
response" which is transmitted by heavily myelinated nerve fibers at speeds up to 45
meters per second- carrying the pain information to the brain from as far as the foot in
as little as 0.1 second (Iggo 1972). Up to one full second can pass before the brain
receives the "slow response" which is carried via nonmyelinated fibers at speeds around
1
meter per second (Iggo 1972). The slow response gives the brain the important detailed
information about the pain's location and type of stimulus that the fast response lacked.
The parts of the brain that are involved with the pain sensation include most of the major
higher brain centers. The higher brain centers can affect the lower level reflex response
(Miller 1967). For example, if one was to pick up an extremely hot cup of tea in an
20
expensive cup, the hand will still jerk away but not drop the cup. This demonstrates that
after the brain perceives pain it must process the pain information on several levels and
then select the appropriate response and act upon it. Since the modifications made to the
SFUS system require the patient to press a button in response to pain (presented later) the
above reaction times should be kept in mind.
1.4.3 PAIN, THERMAL STIMULATION, AND DRUGS
Several studies have been carried out to determine the relationships between power
deposition, tissue temperature, and pain sensation (Hardy 1953; Hardy and Stolwijk 1966;
Hardy et al. 1967). From experimentation done on artificially blackened spots on the
human forehead and on the back of human hands, it is possible to find approximations
relating tissue temperature and the amount of steady state power required to reach the
pain threshold. The experimental setup that Hardy used consisted of an intensity
controlled radiant heat source that was separated from the subjects being tested by a
shutter.
This system exposed the subjects to a step increase in radiant heat exposure
when the shutter was opened. The resulting temperature rise at the skin of the subject
would then approximate an exponential function of time and the radiant intensity. These
studies found that the sensation of pain depends solely on the elevation in the temperature
of the tissue as long as the intensity of the radiation was sufficient to elevate the skin to
a painfiil temperature. This is in contrast to tissue injury which was found to depend on
both the temperature and the duration of heating (Hardy et al. 1953). The dependance
21
of the sensation of pain on the skin temperature can be seen by looking at equation 1
which was found by Buettner and used by Hardy after exposing 6 subjects to different
levels of radiant heat and measuring the corresponding time to reach the pain threshold
in 90 different experiments.
Tg = To + Qk4t
(1)
Where T, = skin temperature at the pain threshold (°C); To = initial skin temperature
before irradiation (°C); Q = intensity of radiation (mcal/sec/cm'': k = an experimentally
determined constant = 0.032; and t = time of exposwe in seconds.
When Hardy used equation 1 to calculate the skin temperature at which the a
given intensity evoked a pain response, he found that a temperature of 45.7 °C with a
standard deviation of 1.7 resulted from all experiments regardless of the length of time
or level of power used which agreed closely with experimentally gathered temperatures.
Since higher intensities of radiation will cause a faster temperature rise on the skin being
exposed, the time rate of change of temperature at the skin surface (dT/dt) is a major
factor in the amount of time that is required for the pain threshold to be reached at a
given temperature; i.e., a fast temperature elevation (from a high intensity of stimulation)
will result in the pain threshold being reached sooner. The temperature of 45.7 ±1.7
°C puts limits on the applicability of equation 1 because a skin temperature of 42 °C
would be out of this range and would therefore not cause pain. However, there is
variance in the pain threshold (T,) over different parts of the body. The heel of the foot
has the highest threshold, and the lower back, buttocks and thighs have the lowest
22
threshold to a painful stimulus (Hardy et al. 1967). The pain threshold will also be lower
in areas where the tissue has been previously damaged. Rapid repetition of stimulation
to the same area will also lower the pain threshold by causing sensitivity to the stimulus
until habituation occurs- after which time the threshold will increase (Hardy et al. 1967).
A pre-existing pain in one area can also cause an increase in the pain threshold in a
different area of the body (Hardy et al. 1967). (This is the rationale behind the practice
of "biting the bullet" during painful procedures such as tooth extraction without the
benefit of anesthesia.) This could be caused by an overload of the nervous system or the
body's production of natural pain suppressors created by the first pain sensation. The
pain threshold can also be increased by a person's attitude, with one study reporting that
20% of subjects tested experienced an increase in the pain threshold by repeating the
phrase "I don't feel the pain" (Hardy et al. 1967). The same study also found that 90%
of subjects tested reported an increase in the pain threshold when placebos were
administered (Hardy et al. 1967). Overall, 30% of all patients in pain can expect pain
reduction from the administration of medically inert substances (Kiff 1974).
More powerful than placebos are analgesics. A drug is classified as an analgesic
if it is able to act on the nervous system in such a way as to reduce or eliminate pain
without producing unconsciousness (Hardy et al. 1967). Analgesics work principally by
blocking the action of peptides which are formed in the extracellular fluid whenever tissue
damage occurs (Kiff 1974). There are numerous factors that must be considered in the
evaluation of analgesics. Some of the important factors are given by (Hardy 1967 et
al.) as:
23
"1. The pain threshold-raising effect;
2. The pain reducing effect;
3. Reducing of pain by suppression of visceral and somatic effects
themselves pain inducing;
4. Alteration of attitude toward pain by promoting feelings of well-being and
freedom from anxiety on the one hand, or depression or excitement on the
other;
5. Sedative action, inducing sleep;
6. Side effects involving nausea, vomiting, and constipation;
7. Toxic action, particularly suppression of respiration, and allergic reactions;
8. Total effect upon the patient, involving questions of whether the patient
"feels better", and whether the analgesic is suited to administration in the
particular patient."
24
CHAPTER 2.0: SYSTEM DESCRIPTION
2.1 EXISTING SYSTEM BEFORE MODIFICATION
The system used to implement this study began as a conunercial diagnostic
ultrasound unit (Octason by Ausonics Pty. Ltd., Sydney, Australia) that was modified
to produce hyperthermia for the treatment of cancer (Hynynen et al. 1987). To induce
elevated temperatures, six high power 1 MHz focussed ultrasound transducers have been
mounted onto the imaging gantry. The focus of each transducer is aimed at a common
Coupling
Tumor
Patient
(degassed water)
Water _
Degassed
Tank
water
PVC Membrane
s
:\
z
^Rotate
X
Figure 1. SFUS System.
Therapy
Transducers (6)
25
point (Figure 1). The gantry possesses five degrees of freedom (X, Y. Z. rotate, and tilt)
but only two dimensional concentric X-Y octagonal scans were used. Gantry motion is
controlled by an Apple He computer (Cupertino, CA) and programmable stepper motor
indexers (Anaheim Automation, Anaheim, CA). Positional feedback potentiometers
installed on the gantry provide the Apple with a locational safety check. The RF voltage
for driving the amplifiers is generated by a frequency generator (Wavetek model 271, San
Diego, CA) and then amplified by a six channel power amplifier (URI Thermics). The
temperatures were monitored using seven sensor manganin-constantan thermocouple
probes. The thermocouple voltages were measured sequentially by a digital voltmeter
(HP 3456A) via a data acquisition system (HP 3497A) and converted to temperature by
the HP computer (HP 9836).
Figure 2 shows a system block diagram for the.
hyperthermia equipment before modification for this work.
T1
•irt*
ft
!
N>
CfQ
PRE-EXISTING SYSTEM
(A
TO
B
CO
Rpr
D
••
OPERATOR
CONTROL
OF
OVERALL
POWER
FUNCTION
GENERATOR
H
POWER
AMPUFIERS
APPLE
COMPUTER
TO SET
SCANNING
PAHERN
ULTRASOUND
TRANSDUCERS
ULTRASOUND
DEUVEREO
TO PATIENT
SCANNING
GANTRY
MOTOR
CONTROLLERS
1
POSITION
MEASUREMENT
(POTENTIOMETERS)
TEMPERATURE
MEASUREMENT
COMPUTER
PATIENT
H
I
TEMPERATURE
SENSORS
IN PATIENT
27
2.2 LIMITATIONS IN THE EXISTING SYSTEM
The SFUS system at the University of Arizona did not originally have any
practical means of controlling the ultrasound power as a function of scan position during
patient treatments. The power field could be increased or decreased as a whole, and the
power of each transducer could be varied as well as the frequency of the transducers. In
addition, if a patient experienced pain during a treatment, the entire scan would have to
be moved away from the area of discomfort or the overall power would have to be
lowered in order to relieve the discomfort. This work was undertaken with two basic
goals in mind.
First, to provide the operator with a robust temperature feedback
controller to modulate the ultrasound power as a fimction of scan position in order to
improve the temperature field. Second, to obtain direct feedback from the patient as to
when pain is experienced and use this information to alter treatment parameters in order
to alleviate the pain.
2.3 NEW SYSTEM DISPLAY
To improve the above mentioned limitations in the University of Arizona SFUS
system many additions and modifications have been made to the system described in
Section 2.1. One of the first additions to the system was a treatment display screen.
MENU
SCAN OF
CENTER
DIRECTION OF
ADJUSTABILITY
SCREEN AREA
INNER
SCAN
100%
PAIN
WINDOW
12
^100%
0-360
LIMIT
iSo
100%
100%
SECTOR LINE
END POINTS
100%
100%
100%
Quit
Window diameter 20 mm
Controller targota
Zoom
ControUer ia off
Redraw ccroon
Reset windows
Move center
100%
Align tumor trace
Enter tumor and TC'o
ikzven tar)get 42.t> C
l/D validaticm
Bntur GSdge Point
GANTRY
LOCATION
OUTER
SCAN
SCAN PATH
HISTORY
29
Figure 3 shows the display screen during a typical treatment. The actual screen has the
benefit of color to distinguish the different elements of the screen from one another.
There are two main parts to Figure 3, the menu to the left and the graphic display area
to the right. Both parts of the display are described below.
The menu lists the different commands that are available to the system operator
during a treatment. A moveable check mark (not shown) to the left of each item in the
menu list is used to indicate which menu item is to be selected. The meanings and
restrictions on the use of the commands is described in detail in the users manual located
in Appendix A.
The following procedure is used to enable the temperatures to be displayed in their
correct locations and to allow the controller to assign thermocouples to the proper sectors:
first, before the start of a treatment a transparency is placed directly on the patient over
the site to be treated. Next, a rough tumor outline is marked, and the insertion sites of
the temperature probes are marked along the axis of the insertions. This information is
combined with length calibration marks on the probes to determine the depth of
penetration. This information is then drawn on the transparency. An alignment mark is
then made on the patient and on the transparency which indicates the center of the scan.
The information on the transparency is then entered into the Vectra computer using the
graphics tablet.
The lines of numbers in the graphical portion of Figure 3 represent thermocouple
temperatures displayed with the decimal point indicating the thermocouple's position
relative to the rest of the scan parameters.
Each temperature probe has seven
30
thermocouples arranged in a line as shown in Figure 3. The color of each temperature
displayed will be blue if the temperature is below the screen target (set by the operator
and described in Appendix A 3.11) and it will be red if above.
Above the decimal
point a small arrow will appear if the current temperature for each thermocouple is
different than the previous temperature for that thermocouple by more than 0.1 °C. The
arrow will be red and point up if the current temperature is hotter than before or blue and
point down if it is colder. The temperature colors and arrows are only for display and
have no other effect on program execution. If a temperature is less than 0 °C or more
than 80 °C, it is assumed to be an invalid temperature reading and only the decimal point
is displayed.
The small solid circle in Figure 3 represents the gantry position. On the actual
screen it is represented as a brilliant yellow ball (know as "Scott's Comet") that leaves
a trail to mark its previous positions. The trail will be white if the gantry is currently in
an outer scan and it will be green during an inner scan. The trail of past gantry positions
will accumulate until the screen is redrawn by moving a sector line or one of the
following menu conunands: REDRAW SCREEN, ZCX)M, MOVE CENTER, or ALIGN
TUMOR TRACE AND TC'S all of which are described in Appendix A. The buildup
of the trail allows the gantry scan path history to be viewed, and allows any gantry drift
to be readily visible. Occasionally, the trail of past gantry positions may become noisy
creating a haze around the scan path. This is due to noise on the electrical lines of the
hospital or from interference from the ultrasound transducers on the position signals. The
noise will have no adverse effects on the treatment as long as the magnitude of the noise
31
does not become excessive in comparison to the size of the scan. A first order low pass
filter has been installed on the signal lines to help alleviate as much of the noise as
possible.
The radial lines in Figure 3 are used to show how the scan field is divided into
eight radially adjustable sectors and four inner radial sectors for a total of twelve
temperature control regions.
The boundary between the inner and outer scan regions
used by the power control (described later) is shown in Figure 3 as a circle that is
concentric with the center of the scan. The solid arrows in Figure 3 were added to
illustrate the direction of adjustability for the radial sectors.
2.4 COMPUTER CONTROL OF TEMPERATURE
Computer control of ultrasound power as a function of scan position would result
in a significant improvement in SFUS treatments. Two methods of control are provided.
The first method is manual control of power for each of the 12 sectors in Figure 3 and
the second method uses a temperature feedback system.
2.4.1 MANUAL POWER CONTROL
If the cursor keys are used to move the menu pointer past the top or bottom of the
menu list (described above), the pointer will move to the perimeter of the gantry display
screen. If the menu pointer is advanced all the way around the perimeter and there is an
32
inner scan, the menu pointer will then advance to the perimeter of the inner scan around
it, and then back to the menu list.
When on the perimeter, the menu pointer will
alternate between a power adjustment for that sector (represented as 100% in Figure 3)
and a sector line end point. Sector lines are used by the program to define independent
areas for power control as a function of gantry position. Whenever the gantry enters a
new sector the ultrasound power will be adjusted to the percentage specified for that
sector (either by manual control or the automatic temperature controller). The power
adjustment can be in two forms depending on the state of the temperature controller. A
percentage of between 0% and 100% with adjustable increments of 10% will be shown
if the temperature controller is OFF (as seen in Figure 3); or. the controller target
temperature will be shown for that particular sector in degrees celsius with adjustable
increments of 0.1° C if the temperature controller is ON (the 100% shown in Figure 3
would be replaced by the target temperatures). The " +" andkeys on the right of the
keyboard can then be used to either move the sector lines radially cw or ccw in 5 degree
increments or to adjust the power or target temperature up or down as described above
depending on which one (sector line end point or power/target temperature adjustment)
the menu pointer is on when the keys are pressed. The program puts two restrictions on
sector line movement: the first is that one sector line cannot cross over another line, and
the second is that the first and last sector lines cannot cross over the 0-360 degree line
which the pain program defines as residing at the middle right portion of the screen.
33
2.4.2 TEMPERATURE FEEDBACK SYSTEM.
This section will describe how the uniformity of the temperature field in the
tumor is improved while relieving some of the operator's responsibilities in temperature
monitoring and power adjustment by use of a modified implementation of the controller
described by Lin et al. 1990, Lin 1990.
The software developed for this section is essentially two dimensional as is the
present scanning pattern utilized by the SFUS system. The patients however are not two
dimensional but a sub-class of their tumors can be considered as being two dimensional;
such as breast tumors. Breast tumors and the surrounding areas are of somewhat uniform
thickness and require temperature probes to be placed parallel to the plane of the scan.
The assumption was made that during a successful treatment limited thermal
coupling would generally exist between the sectors that were shown in Figure 3. That
is, when the temperature field is close to being uniform, the thermal conduction between
adjacent regions is small and can therefore be neglected. If the controller works well this
assumption would still be valid even under conditions of highly nonuniform blood flow
because a good controller would control the power in such a way as to improve the
temperature field until the effects of thermal coupling are small and can be neglected.
The assumption of limited thermal coupling then allowed each sector to be given an
independent modified version of the single input single output (SISO) bang-bang/PID
34
controller developed and analyzed by Lin et al. 1990, Lin 1990. A block diagram of this
controller can be seen in Figure 4.
TARQET
BWOR /* .>
I COMPARATOR.|^
POWBl
, ULTBASOUND
,
AMPUi^ ^^TRANStXJCSl'^
TUUOR
BAIIQ4ANQ
MAXIMUM
TB4PERAnjRE
Figure 4.
AUTO REGRESSIVE
MOVING AVERAGE
MEASURED
TEMPERATURES
Controller Block Diagram.
The analog form of the controller is given in equation 2.
•(t) +
0
t)
—
f eit) dt+x. de{
t,- Jt,dt
if r.^r.argec
(2)
Where
= target temperature; T, and T; are the upper and lower limits for the bang-
bang controller such that when the measured temperature is between T, and T, the PID
portion of the controller is used; Paa, = upper limit for the amount of power that can be
delivered (zero is the lower limit for power that can be delivered); P = power selected
by the controller; e(t) = difference between the measured and target temperatures; K^,
35
Ti, and Ti, are the proportional gain, integral or reset time, and derivative time,
respectively.
The PID section of the controller is written in digital form by using a backward
difference approximation for the derivative, a trapezoidal numerical integration for the
integral, a sampling time of 30 seconds, and a gain of 0.4 (Lin et al. 1990, Lin 1990)
(equation 3).
Dnax
p J
-n
(3)
100
0
In equation 3, n = the current time step; n-1 = the previous time step; P„ =
power at time step n; e„ = difference between the measured and target temperatures at
time step n;
= the sum of all the errors from n = 0 to the current time step.
The main features of this control law are that it gives fast rise times when the
temperatures are low, fast drop off times when the temperatures are high, and an
accurate PID controller when the error is small (Lin et al. 1990, Lin 1990). To reduce
the effects of noise and seasoning, an auto regressive moving average (ARMA) was
employed to express the controller input temperature-in terms of a linear combination of
the current and previous measured temperatures. The weighting factors for each of the
temperatures in the temperature history were experimentally determined (by Lin et al.
1990, Lin 1990) for use on the same system as the one used for this smdy. Equation 4
shows the relation used.
2'^=o. 3 5 r^+0.3
+0.2 r^.2+0.15
(4)
One property of a controller that contains an integrator is that it will try to drive
the integrand (which is the error signal) to zero. This can be a potential problem when
the error stays either above or below the saturation level of the input for an extended
period of time (integrator windup), thus accumulating a large area under the error curve.
Then, if the error switches to the "other side of the saturation level, the controller would
need to counteract the large area accumulated by the integrator previously. This leads
to large overshoot and errors (Franklin et al. 1991, Middleton and Goodwin 1990).
Integrator windup is avoided by integrating the error (or summing the error when using
discrete time) only when the error is between the bounds of the bang-bang portion of the
controller (between Ti and Tj), thus reducing the area under the error curve yielding and
a faster response time.
The anti-integrator windup is utilized throughout a treatment
whenever the error for a sector is outside of the bounds of the PID section (T^ro. > Tj
or T^ < T,).
37
One drawback of using a SISO controller analogous to Lin's is that it required
preselected thermocouples to be used as input tor each sector that was to be controlled.
In order to let the computer select the input to the controller for each region in real time,
and to accommodate multiple thermocouples in each region, the thermocouple in each
sector with the highest temperature according to equation 4 (ARMA) is used for the input
to the controller for that sector. The maximum is selected in order to prevent run away
temperatures and to help prevent the controller from saturation caused by trying to raise
the temperature of a control thermocouple that was not significantly affected by the
applied power.
A potential problem with a PID controUer can occur during the transition from
manual to computer control if the error is close to zero. The controller would begin
operation with a small error, and since this would be the first time through the control
law, the integral of the error would also be small. This condition would yield a low
power setting. If the power was at a high level to achieve the previous temperatures, the
resulting low power setting from the control law would produce a dip (or bump) in the
temperatures. The problem is more severe if the control temperature is above the target
temperature but still requires attention when the control temperature is below the target
temperature. In order to achieve a bumpless transition to automatic control (see Astrom
38
and Wittenmark 1984 or Middleton and Goodwin 1990), the sum of the error term in
equation 3 was initialized according to equation 5 which was derived from the PID
section of formula 3 by setting n = 0 and solving for the summation term.
Eo ^ _ 200Pq + 8©_T ~ 88SQ
-® J
30
(5)
Equation 5 would give the same power level after transition to the control law as
was required before transition, thus creating the bumpless transfer.
2.4.3 SOFTWARE STRUCTURE
The program to implement this work was written in Microsoft Basic version 6.0.
Microsoft Basic offered the advantage of being able to create code that could be compiled
and run more quickly while still maintaining a somewhat universal understanding among
fellow programmers. The source code contains comments detailing the function of each
line. The variable declarations contain comments that explain the ptirpose, meaning, or
range of acceptable values for all variables used in the program.
39
2.4.4 PROGRAM LOOP
The program written to implement this work is structured with three levels of
code. These levels of code can be categorized into the program loop, main program, and
the error handler. Before the
program enters the program
INITIALIZE PROGRAM
loop (Figure 5), ail of the
variables are initialized, the
CALL MAIN PROGRAM -
data array are declared, and
, NOlNPUr :
CHECK FOR KEYBOARD 1
INPIJT
all of the communication lines
with the I/O cards, graphics
^J
INPUT
tablet are configured, and the
CALL KEYBOARD
INTERRUPTER
patient data files are opened.
Figure 5.
r
^
Program Loop.
The program loop was written
so that it is possible to acquire user keyboard input while still executing the main
program. This was done by first calling the main program and then checking to see if
there has been any keyboard input. If keyboard input occurs, the keyboard interrupter
section is called. The interrupter section was written in a manner to avoid requiring the
computer to wait with statements like INPUT, which demand 100% of the computers
40
attention until input is received. By using the INKEY$ command, which checks the
keyboard and then continues, it became possible to write keyboard data entry code that
would both interrupt key presses and caU the main program without any noticeable delay
to the user.
2.4.5 MAIN PROGRAM
The main program contains the temperature controller, pain button algorithm
(described later), all data input and output, and screen graphics routines. The entire
section is structured so that when called, there are no delays or loops in the code. This
allows the program loop to operate as quickly as possible. The methods used also
streamline program execution by allowing this section to be called at a faster rate
increasing the data I/O sample rate. Streamlining was accomplished by an extensive
application of data flags to keep track of the current state of all major operations. To
further enhance the optimization of this section,
decision blocks are used to skip
potentially time-consuming operations like data I/O when they are not needed.
A
complete block diagram of this section can be seen in Figure 9.
The main program also acquires and displays temperature information over the
IEEE bus.
Since the Vectra computer is configured as a peripheral on the bus it is
therefore required to wait for data to come across the bus; verses commanding that it be
sent. There are two basic methods for detecting when data is being sent. The first would
have the program continuously check the state of the bus for data. This method is
41
SEND CXJT SAFETY SIGNAL
I
I
I
AQUIRE NEW GANTRY
3AN1
POSITION
NO
HAVE SCAN PARAMETBtS CHANGED
YES
yes !
HAS PAIN BUTTON BSN PRESSED
AUaORITHM
NO
I
HAS SCANNING APPLE COMPUTER
0^ ^
SET A NEW CENTER
I
NO
SCANNING APPLE IN AN INNER OR
OUTER SCAN
I
DETERMINE THE SECTOR THE GANTRY ISI
CURRENTLY IN
I
T
I
I
I
HAVE THERE BEEN ANY PAIN BUTTON PRESSES
YES
DETERMINE IFTHE GANTRY IS IN A PAiN WINDOW
NO
I
DOES THE POWBR NSD ADJUSTMENT
YES
ADJUST POWER
ARE THERE TEMPBMTURES TO DISPLAY
AND DO THEY NEH5 UPDATING
T
END OF MAIN PROGRAM
Figure 6.
Main Program Block Diagram.
ID SCAN
PARAMETERS
42
extremely slow, taking up to one microsecond per check, which would effectively cut the
program data sample rate in half. The other method would set up a program interrupt
to cause the program to junq) to a 'read data bus' routine whenever data is present. Since
the later is done at a machine code level it is far iaster and does not use as much valuable
CPU time when not needed. Because of the speed advantage, the interrupt method is
used. (It should be noted that this interrupt can be rendered inoperable by an incorrect
installation of device drivers in the CONFIG.SYS boot file of the Vectra computer.
Device drivers for components like a mouse MUST be installed before the device driver
for the A-488 card.) Data is sent over the bus two temperature probes at a time. Each
set of probe data is stored in a data array until the count on the number of probe data sent
matches the number that tiie pain program is expecting at which time a flag is set to cause
the main program to update the temperature display and call the temperature controller
(if it is activated).
43
2.4.6 ERROR TRAP
BtROR
Since a program crash would
IEEE 488
BUS ERROR
YES
- RE-WmALIZE BUS :
have adverse effects on both the
RESUME PROGRAM 1
DOS ERROR
effectiveness and more importantly,
the patient's safety, an error handler
becomes a necessity.
During the
MAXByUi NUUBERi
ORBWORS
EXCEEDED
NO
- DISPLAY5?R0fl
YES
WATT FORI
488 BUS DATA
execution of an application specific Figure 7.
•
Error Handler Block Diagram.
program over 4000 lines long, there are many types of errors which can occur which do
not warrant shutting down the entire treatment system. Examples of non- treatment
threatening errors would include trying to locate graphics outside of the acceptable range,
a bad reading from one of the 1/0 cards, invalid user input, or a program operator using
the program in situations that have not yet been conceived. While the first three above
mentioned errors have been accounted for, accounting for the third is not possible. The
error handler outlined below tries to separate critical errors from non-critical ones.
The error handler (Figure 7) for the program is initialized by setting a program
interrupt to jump to a special section of code to deal with an error when it occurs.
Without the interrupt the program would stop execution when an error is detected. The
code that the program jumps to first checks for the most likely source of the error; the
A-488 bus. If the bus is the source of the error, the error handler will re-initialize the
bus and then resume program execution. If the error was not a bus enor, it is assumed
44
that there may have been a pain program error.
If this is the case, a counter is
incremented and program execution is resumed in the hope that the error which occurred
will not occur again. If the number of errors exceeds 20 (an arbitrary number) there
must be a more serious flaw in the code and all program activities are halted except for
the reading of the A-488 bus to prevent the HP temperature computer from hanging up
and the power is set to 0%. At this point the interrupt method for reading the bus is no
longer available so the first method of bus control is used.
2.4.7 HARDWARE FOR COMPUTER CONTROL OF TEMPERATURE
Several hardware additions to the system shown in Figure 2 were made for this
work. These additions are shown in Figure 8. The most significant addition in Figure
8 is a HP Vectra QS/20 (Hewlett Packard,Cunnyvale, CA) 386 computer. The 386
computer was integrated into the old system by tapping into the gantry positional
information generated by the position measurement potentiometers and scan information
from the Apple scanning computer. The temperamres Irom the temperature measurement
computer are sent to the Vectra QS/20 over an IEEE-488 bus between the computers.
Thermocouple spatial information is entered into the 386 by the use of a digitizing
graphics tablet (Summagraphics Corporation, Seymour, CT). The 386 computer can
control ultrasound power by commanding a duty cycle controller box which intercepts the
function generator ouQ)ut before the signal goes to the power amplifiers.
PRE-EXISTING SYSTEM
OPERATOR
CONTROL
OF
OVERALL
POWER
n
POWER
AMPUFIERS
APPLE
COMPUTER
TO SET
SCANNING
PATTERN
FUNCTION
GENERATOR
ULTRASOUND
TRANSDUCERS
SCANNING
GANTRY
MOTOR
CONTROLLERS
ULTRASOUND
DEUVERED
TO PATIENT
I
PATIENT
POSITION
MEASUREMENT
(POTENTIOMETERS)
SCAN
INFORMATION
TEMPERATURE
SENSORS
ADDITIONS TO SYSTEM
DUTY CYCLE
CONTROLLER
HARDWARE
SAFETY
SHUT-OFF
OPERATOR
POWER
DECREASE
FIBER
PATIENT
PAIN
BUTTON
OPTIC
CABLE
HPV-20 386
COMPUTER
GRAPHICS
TABLET
TEMPERATURE
MEASUREMENT
COMPUTER
46
The duty cycle controller box has six independent channels each capable of
imposing a duty cycle of between 0% and 100%. At the current stage of development,
all six of the channels are controlled in unison. The duty cycle controller operates by
ANDing three signals together to form the output for each channel. The first signal is
formed from a 100 Hz spike trigger that was generated by a free running 555 timer
(Figure 9). The trigger signal is then distributed to six 555 timers that are configured to
generate one pulse for each trigger spike. The duration of the pulses from each mono
shot timer is determined by analog voltages generated by the Data Translation 2811 D/A
card installed in the Vectra computer. As long as the duration of the pulses are less than
the length of time between the trigger signal pulses, a 100 Hz duty cycle signal is
generated. Thus, if the length of the mono shot timer pulses are three quarters of the
length between the trigger pulses," a 75% duty cycle is generated (Figure 9).
1(X3Hz TRIGGER SIGNAL
Figure 9.
The Anatomy of the Trigger Signal.
DU7YCYCLE (XDNTROLLED SIGNAL
47
The duty cycle signal is then the first signal that is fed into the AND gate (Figure
10). The second signal is the input signal to the box which must be a TTL compatible
signal of any frequency greater than about ICXX) HZ (Figure 10). The 1000 Hz restriction
is imposed because the signal being controlled must have a frequency at least 10 times
greater than the frequency of the duty cycle signal for the scheme to be effective. The
hardware safety shut-off is the third signal into the AND gate. The safety signal is the
output of a CD4538 timer (Figure 10) which will remain high as long as the program on
the Vectra computer is running. This is accomplished by having the main program send
one pulse of a square wave TTL signal that is generated by the Data translation 2801
digital I/O card. This safety signal is used to continuously trigger the CD4538 timer.
Therefore, if the safety signal stops for any reason (such as a computer crash or program
bug), the output of the safety timer will go low, disabling the output of the AND gates
8L
DUTYCYCLE SIGNAL
&
1MHz TTL SIGNAL
DUTYCYCLE CONTROLLED SIGNAL
8l
" SAFTETY SIGNAL
Figure 10. The Anatomy of the Dutycycle Output.
48
and effectively shutting off power to the transducers. Appendix C shows a complete
schematic for the duty cycle box.
2.5 INTERACTIVE PAIN BUTTON DESCRIPTION
In order to identify the area of patient discomfort, an optically isolated button was
constructed which is referred to as "the pain button" (Anhalt 1991). When in the hand
of a patient, the pain button allows the program monitoring it to get direct input firom the
patient as to the approximate location of any pain. This bypasses the slow and inaccurate
verbal identification procedure that existed previously. The pain button is most effective
if the pain appears and disappears quickly in the same location with the same period as
the repetition of the gantry scanning pattern. This type of cyclical pain can occur when
ultrasound strikes a bone or a nerve and will usually be sharp pain in nature. It is this
type of pain that the focus of this portion of this work.
In order to correlate the position of the gantry at the time of the pain the patient
is asked to press a pain button that is monitored by the Vectra computer. Throughout the
rest of the treatment, the program adjusts the duty cycle (power) of the ultrasound
whenever the gantry location is close to a stored pain button press location.
49
2.5.1 PAIN BUTTON ALGORITHM
The goal of ±e following algorithm is to alleviate the pain while administering
as much power to the patient as possible. During the development of this algorithm any
decisions or assumptions that had to be made were done in such a way that any potential
errors would be on the side of patient safety, in contrast to being on the side of best
temperatures or most ultrasound power delivered. To this end. most button presses lower
the power. A method was required to deal with multiple pain locations and to account
for the fact that the location of the gantry at the time of the press will move past the
pain's true location due to a patient's finite reaction time. Patient reaction time is an
uncertain quantity due to medications, slow reflexes, or poor physical/mental patient
condition as discussed previously.
The algorithm works by searching for the correct delay time of the patients
reactions to pain and by finding the correct size and location of the area that is causing
the pain. The program then lowers the applied power in that area to a level that is as
therapeutic as possible while not being intolerably painful. This algorithm is described
herein for one pain location and thus one series of windows. However, the program can
apply the algorithm to multiple locations simultaneously. The limit on the total number
of wiodows at any one time is 200 (the previous version of the program had a limit of
300 pain button presses throughout the entire treatment). If the limit is exceeded, the
program will beep repetitively and display a message stating that no more presses will be
accepted.
50
All displayed window powers (and all powers discussed below) are a percentage
of the maximum absolute power of the sector that they are in. For example, if a window
with a power setting of 50% is in a
sector with 80% power, the absolute
Sector at 80%
power
power would be 40% of the maximum
power.
Absolute
power
40%
If a window area overlaps
Window at 50%
power y
/
Sector
fln®
Absolute
power
30%
two sectors of different power levels,
the absolute power when the gantry
X^Scan
Sector at 60%
power
position is inside the window will
change accordingly when the gantry
^
Figure 11. Window and Sector Power.
crossed the sector boundary (Figure 11).
The pain button algorithm was limited to three windows in any one series based
on the basic assumption that no pain will cover an area larger than two window diameters
and the patient delay time will not be greater than the time required for the gantry to
travel through three windows. In addition, adding more than three windows to a series
vastly increases the complexity of the algorithm. The algorithm developed can be broken
down into the following 3 simple to follow and understandable cases which try to both
(a) determine the unknown delay time and (b) reduce pains that cover areas larger than
one window. The logic of this algorithm is based on an "ideal patient" who has no
variability in his/her delay time at any given location. However, variations in delay time
and in the size of the painful area between sites is accounted for by keeping each series
of windows independent ^om one another.
51
CASE 1: Presses in the first window.
SITUATION 1.1; Single Window.
STEP 1.1.1) When the pain button is pressed in an area where no previous windows
exist, a window of operator selected diameter and 0% power is established behind the
gantry position at the time of the press
(Figure 12). This window is established
in order to attempt to alleviate the pain
PT9«#
Location
the ultrasound has caused and to account
for the delay time between the position of
Window 1
the gantry when the pain was felt and the
position of the gantry when the button
Figure 12. First Window.
was pressed. A typical scan speed of 40
mm/sec and the default window diameter of 20 mm allows for a maximum of about 0.5
seconds for the delay time of the patient. Future presses in this window control power
in this window.
STEP 1.1.2) Once a window is established, the program will search for the magnitude
of power that can be delivered inside this window without causing pain to the patient, i.e.
no further pain button presses are made within the window.
This is done by
52
automatically increasing the percentage of window power in the controlled window by
10% with each pass through the window up to a maximum value of 90%.
STEP 1.1.3) If the pain button is pressed a second time when the gantry is inside of the
controlling window, the assumption is made that the power has surpassed the patients
tolerance in the window being controlled. The power will then be lowered in the
controlled window by 10% and then be held there unless there is another press in the
controlling window, in which case the power in the controlled window would be lowered
an additional 10% and held at this level. This process can be repeated until 0% power
is reached. The logic followed for future presses in this window when it's power is at
0% is covered below.
SITUATION 1.2: Double Windows.
STEP 1.2.1) The patient could at
sometime press the pain button in the
Press
Location
controlling window while the power is
at 0% in the controlled window. This
could occur if the patient delay time
Window 2
was greater than 0.5 seconds or the
area that caused pain is larger than
one window diameter.
When this
Figure 13. Second Window.
Window 1
53
h^jpens, a second window with 0% power is established behind the first whose power
is controlled by button presses in the first window. This is shown in Figure 13 where
the arrows below the windows indicate that presses in window 1 are controlling the power
in window 2.
STEP 1.2.2) Windows 1 and 2 will start increasing in power in the same manner as
described in steps 1.1.2 and 1.1.3 in
order to determine the patient's tolerance.
Both window powers are grouped together
at the same power because the patient's
\ Scan
Path
\
Possible Pain
Location
Ik
Press
Location
delay time might be less than the time
required for the gantry to move through
Window2
Window 1
one window and the pain area might
Figure 14. Possible Pain Location.
overlap both windows as shown in Figure
14. This is a compromise in the conservative direction in the face of uncertainty about
the delay time of the patient and the size of the pain region. It should be noted that
ftiture algorithms should try to more thoroughly deal with this situation.
54
SITUATION 1.3: Triple Windows.
STEP 1.3.1) If the patient delay time was
longer that the time
required for the
Scan
Press
Location
gantry to move two window diameters
\
causing the pain to persist with two
windows (indicated by a press when the
/
Window 3
Window 2
Window 1
gi;antry is in the first window while the
power in the first and second windows are
Fig:ure 15. Third Window Location.
0%), a third window with 0% power is
established behind the second (Figure 15).
STEP 1.3.2) The power in window 3 then increases in the same manner as described in
step 1.1.2 and 1.1.3. The arrows between windows 1 and 3 in Figure 15 are to show
that the power in window 3 is being controlled by presses in window 1 in the same
manner as described before. The power in windows 1 and 2 are reset back to 100%
since they were ineffective. It is assumed here that the patient delay time is being
searched for because the search for the size of the pain area was not effective. This is
the one exception to the general rule of having pain button presses lower the ultrasound
power. The exception was made because three windows in a row with low power
settings would prevent major portions of the target volume firom being heated.
55
STEP 1.3.3) Assuming the patient delay time is known and that the pain covered an area
larger than one window diameter (as indicated by a press in the first window while the
power in the third window is 0%), the power in the second window is again reduced to
0% in order to cover the area of pain that the third window did not cover. The second
and third window powers are then controlled by presses in the first window in the same
manner as described for step 1.1.2 and 1.1.3.
SITU.^TION 1.4: Complications in the above steps.
STEP 1.4.1) The first complication occurs
Scan
in step 1.2.1 when there is a pre-existing
Path
WMawfrem
dlffsrant series
window of a different series and 0%
power behind the first. Since there would
Window 2
be no benefit in establishing a window at
Window 1
a location that is already at 0% power, the
"second" window is established behind the
pre-existing window (Figure 16).
Figure 16. Second Window with Pre­
existing Window.
The
arrows between windows 1 and 2 in Figure 16 indicate that presses in window 1 are
controlling the power in both windows. The power setting of windows 1 and 2 shown
in Figure 16 are set as was described in step 1.2.1. If the pre-existing window from the
different series has a power setting greater than 0%, then the "second" window is
established on top of the location of that pre-existing window. This situation is then the
56
same as step 1.2.1. It is then possible for
the gantry location to be inside of more
than one window (i.e. the area where
Scan
\ Path
Overiap
«
Gantry
-^Location
windows are overlapping in Figure 17).
When this happens, the gantry
is
Windows from Different
Series
considered to be in whichever window has
a lower power setting. This conforms to
Figure 17. Overlapping Windows.
the overall conservative approach used in setting power levels.
STEP 1.4.2) The second complication occurs from a similar situation that can occur
when the program is establishing window 3 in step 1.3.1. The power setting of windows
1 and 2 shown in Figure 18 are set as was
described in step 1.3.1.
The program
looks behind the second window to see if
Windows
\/
a window from a different series and 0%
Press
Lx)catk)n
power has already been established there.
If such a window is found the third
Pre-exMng
window
Wlndow2
Window 1
window is established behind it (Figure
18) for the same reasons given in step
Figure 18.
Third Window with Pre­
1.4.1. It is then possible for the gantry existing Window.
location to be inside of more than one window (i.e. the area where windows are
57
overlapping in Figure 17). When this happens, the gantry is again considered to be in
whichever window has a lower power setting.
CASE 2: Presses in window 2
STEP 2.1) If a button press occurs in a window whose power is being controlled by
presses in a different window it starts a new series. Pain button presses can then arise
in one of two different windows. The
first window that the press could occur
inside of is a "window 2" that is
New press
Loaction
controlled by presses in a "window 1" as
shown
in
Figure
19.
Such
Window
"windows 2's"
can arise
Or^nal
Press
Location
from
the
Old
Window 2
\^Okl
Window 1
conditions that created either Figures 15
or 17.
The logic controlling the new
Figure 19.
Part 1.
Press in Existing Window
series of windows is started from step
1.1.2 where the power in the new window is controlled by presses in the old window 2
as shown by the arrows between the windows in Figure 19. The power settings of the
windows from the old series are not altered.
58
Scan
Path
OLD
Press
Location
New y
Window
Old
Window 3
New press
Location
Old }
Window 2
\ Old
Window 1
Figure 20. Press in Existing Window - Part 2.
STEP 2.2) The next possibility is that the a pain button press occurs in a "window 2"
of Figure 16 or 18. In this situation the new window would be established behind the
old window 3 (Figure 20). The new window would then be controlled by presses in the
old window 2 as shown by the arrows in Figure 20.
59
CASE 3: Press in window 3.
STEP 3.1) If a press occurs in a "window 3" of Figure 15 or 18, it is handled in the
same manner as shown in Figure 19 with old window 2 replaced by old window 3. The
power in the new window is then started from 0% and increases in the same way as
described in steps 1.1.2 and 1.2.3.
Figures 21-23 show a flow chart of the complete algorithm.
START
PRESS?
YES
ESTABUSHWOCOWl
COKTINUE
INCAQUEPCW^ll
1 BY10ftUPT090%
aiNl£S3HBI3| AFTER
EACH PASS THROUGH 1
NO
PfiESSINI?
powBttum?
YES
LCMBtPOV^MI
SYIMAMOHCXD
0QSTM3WMOCW
BBflOl?
YES
ESrABLBHMNXlWS
pcMiBtcoirnnuB)
BYPRESSESMI
WBBOSTWaWMDOW
NO
ESTABUSHWIOOWS
POMStOONTROUH)
BYPRESSESM1
2 1
0*
0*
0% A 0* /'v 0% .
CONTTNUE
mCREASEPOIMe)M
1 AZBrmuPTom
CUNl£SSfeD)itf^
EACHPASSTWOUOHZ
UM9P0IVGRM1i2
SyiO%MOHCXI>
NO
PflESSMI?
YES
P0WetM20%7
YES
PPESSIN2?
YES
START NEWSBSES
0%
NO
y ant V om
t > wn '
B
Figure 21. Pain Button Algorithm - Part 1.
EXmiNQ WINDOW
DEHir4D27
I no
I YES
ESTAHUSH WINDOW 3
POWER CONTROLLED
BY PRESSES IN 1
ESTABUSH WINDOW 3
POWER CONTROLLED
BY PRESSESINI
PRSXtSTlNQ WINDOW cm
3
OK
1
I 6* A ' A, *** A '®* /
2
CONTINUE
INCREASC POWER IN
3BY10H UPTOWH
NO
(UNLESS HEU>) AFTER
PRESS IN 3
PRESS IN 17
EACHPASSTRROUOHa
YES
j YES
NO
START HEW SERIES
POWER IN 30%?
LOWER POV^INd
BYIOHANDHOIB
NO
YES
/ OH •
\ MMt
PRESS IN 27
YCa
RESET POWER IN
WINDOW 2 TO OS
3
2
1
• [ 0%.l 0%/^ /
r- r , •/'
( c )
( 0 )
otDj^'
V ou)t J cioi \
A
A "
/
62
LU
ii
Figure 23. Pain Button Algorithm - Part 3.
63
2.5.2 PAIN BUTTON OPERATION
When the pain button is pressed (signified by a BEEP from the computer), a
circle will be drawn on the screen behind the location of gantry at the time of the press
creating a pain window (shown in Figure 3 as a circle just below the gantry position).
The pain button is debounced both in hardware and software. The hardware debouncing
consists of a capacitor across the switch contacts. The software debouncing requires that
one second must elapse between button presses.
The diameter in millimeters of a pain window is selected by the operator.
The physical size of the window on the screen is dependant on the screen scale which is
adjusted manually by zooming the size of the screen and automatically by the program.
When first created, the circle defining the window will be red. A circle that is all red
signifies that the ultrasound power will be 0% of the sector power whenever the gantry
position is inside. As the gantry passes in to and out of the pain window, and more pain
button press occur inside of the window, the percentage of power will change both up and
down. How the power changes in response to these events is dictated by the pain
algorithm detailed previously.
As the percentage power changes, the circle will change fi"om red to white. The
amount of white on the circle is proportional to the percentage of power inside of the
window it is defining. This allows the percentage of power inside of a window to be
read like the hand of a clock. Thus, a circle that is half red and half white would be at
64
50% power. The power percentages are in 10% increments which translates to 36
angular degrees on the circle per 10% change in power.
2.5.3 PAIN BUTTON HARDWARE
The pain button is a momentary contact, normally open switch mounted atop two
small batteries. When the switch is closed by the patient, current is sent along a cable
into the duty cycle box where it lights a LED. A photo detector senses when the LED
is illuminated. The electrical isolation provides the patient with a safe method for
interfacing with the hardware. The output of the photo diode is amplified to TTL levels
and monitored by the digital input of the Vectra's 2801 1/0 card. Appendix C contains
a complete schematic of the pain button along with all of the appropriate pin outs.
65
CHAPTER 3.0: EXPERIMENTS / RESULTS
3.1 PROGRAM VERIFICATION
The correct identification of gantry scan parameters is critical to the proper
operation of the program. Table 1 lists the verifications that were made of the scan
identifications that occur in a treatment.
Automatic scan identification correctly determines the scan size and gantry speed
Correct determination of inner/outer scan throughout the treatment
Screen automatically zooms when gantry bumps the edge of the graphics display
The first scan center aligns the ^umor trace and thermocouple locations
Table 1. System/Program Checks for Scan Parameter Identification.
A graphical display of treatment parameters is only useful if it contains accurate
information. To this end. Table 2 contains the checks made to ensure that the display of
the scan information is correct.
66
The displayed position of the gantry location is correct relative to all other scan
information
Inner sectors 8-12 are only displayed if an inner scan exists
Movement of the center of the scan is correctly displayed
Sector lines are not allowed to overlap
Thermocouple temperatures that are out of the screen scale are not displayed unless
they are selected for use by the controller
Thermocouples are displayed in the correct locations
Thermocouple colors and arrows are displayed correctly
Thermocouple temperatures are only displayed when in the valid range of values
Tumor trace data filters only allow entry and the display of appropriate data
Trace is displayed with the correct scale regardless of how the screen is zoomed
Table 2.
Display Checks from Dog Trials.
Table 3 lists the items that were verified in order to ensure the correct entry
and use of thermocouple information because the temperature data is vital in the use
of the temperature controller.
Individual target temperatures for temperature controller are used for correct sector
Thermocouples are assigned to the correct sectors
The thermocouple with the highest ARMA is correctly identified for each sector
Thermocouples that are disabled by the user are not displayed on the screen and
are ignored by the controller
Table 3. Thermocouple Information from Dog Trials.
67
Table 4 lists the checks required to ensure that the ultrasound power is adjusted
as specified by the program.
Sector lines act as power boundaries
Manual adjustment of sector power sets power for the correct sector
Power selections made by the temperature controller for each sector changes the
power in the correct sector
The power in sectors with no thermocouples is not adjusted by the temperature
controller
Table 4.
Power Checks from Dog Trials.
3.2 EXPERIMENTS
3.2.1 DOG EXPERIMENTS
The initial verification of the hardware and software constructed for this work was
accomplished during the treatment of 13 patient dogs. The patient dogs were selected by
local veterinarians as candidates for anti-cancer treatment and then referred to the Cancer
center for evaluation. The dogs were treated with SFUS hyperthermia if they were
considered to be treatable by this modality. These treatments demonstrated that the
constraction and implementation of the hardware used was correct and that no program
or system errors were present. The dogs trials also validated the correctoess of new
68
features outlined below and explained in the users manual located in Appendix A.
Finally, they provided invaluable experience in the operation of the SFUS system used
at the Cancer center.
3.2.2 PATIENT TREATMENT
After the correcmess of the system was verified it was used for the treatment of
a patient. The patient was selected iirom those normally treated with SFUS at the
University of Arizona. The criteria used for patient selection was the size and location
of the cancer being treated. In the case of the patient data presented below, the cancer
covered a large area and was located on the chest wall with a more or less uniform
thickness. The patient provided the basis for continued study with the system developed.
3.3 TEMPERATURE CONTROLLER
The dog treatments discussed above also served to analyze an initial PI controller
with successful results. The protocol followed for the use of the controller was: During
a patient treatment the manual power control is initially used to adjust the overall power
until the desired temperature is achieved or, the patient's pain tolerance is reached.
When the overall power has not been altered for approximately fifteen minutes and the
temperatures trends in the treatment area have reached a steady state condition, the sector
lines are manually adjusted to separate hot and cold areas of the scan into separate
69
sectors. Any thermocouples that the operator does not which to be used for control are
deactivated at this time as per Appendix A 3.12. The controller is then activated
concurrently with an increase in overall power of about 10% in order to give the
controller room to operate. The controller is then permitted to control the power in each
sector that has thermocouple representation. Power changes to all sectors, as determined
by the PI portion of the controller, occur with each reading of the temperature field at
30 second intervals. The power in sectors without thermocouples is not altered.
As a result of the dog treatments and after subsequent evaluation of everything
learned, the PI controller was replaced with the PID controller described previously
which has been extensively evaluated for stability and response characteristics by Lin et
al. 1990 and by Lin 1990.
(The addition of a derivative term to a PI controller will
serve to improve the stability of a system and increase damping [Franklin et al. 1991].)
The primary modifications to Lin's PID controller include the bumpless transfer from
manual to automatic control, computer selection of the maximum ARMA temperature for
input into the controller, and manually adjustable sectors that assign the thermocouples
to different regions.
3.4 TEMPERATURE CONTROLLER RESULTS
Figure 24 shows temperature plot for all thermocouples in the mmor during a
successful dog treatment which used the PI controller. The controller was activated at
58 minutes and deactivated at 77 minuets.
CONTROLLER ON
CONTROLLER ON
49.0
48.0
47.0
o
U
46.0
IT
3
45.0
£
44.0
m
h
43.0
Q.
'I If
' 'I
, IV ./ /.\ I \ i V •
I /II
//'
/vl
1/^
I
1^
'
•
' •• •
I'
I « V'
I \ \
'v
\• I
!
*•»
/
••
j.\ \ > 'J ^
42.0
] \.
•/
\
\'v /.^v. , \
•7 ' - -
41.0
40.0
40
V.
\
I
50
70
60
TIME(min)
80
90
71
The PID controller has been used on one patient for two consecutive treatments
with results given below. More clinical results will be obtained later after a faster 486
computer is tested and replaces the Vectra 386 computer.
An example of the PID controller used on a patient is shown in Figure 25 (the
controller was turned on at 25 min). In this treatment the transition to control from
manual operation did not use the error initialization described above (bumpless transfer),
and the resulting dip is seen in Figure 25.
44.5
lU
o
lU
cc
44-
CONTROLLER ON
J
42.5 H
4220
25
30
35
40
46
50
55
TIME (min)
Figure 25. Successful use of the PID Controller (With Dip).
60
66
72
The next treatment to the same site of the same patient in Figure 25 is depicted
in Figure 26 with the addition of the error initialization to the controller (equation 5),
creating a smooth transition to computer control. Figure 26 also shows improvement in
the unifonnity of the temperature field when the controller is activated with a target
temperature of 43.0 °C and a time of 39 minutes. Power previously to the activation of
the controller was at 100% in all regions in this treatment.
48
CONTROILERON
47
lU
ac
Ul 3
ot
giS
%s
til
46-]/
45-
44-
43H
42
22
26
30
34
TlME(min)
38l42
T.^
Figure 26. Dipless Transition to PID Control.
46
T3
50
54
73
The unifonnity of the temperature field can be measured by the variance of all of
the temperature probes in the tumor. Figure 27 shows that the variance dropped during
the second treatment from approximately 8° C to 2° C, which is a significant
improvement.
1211 -
CONnK)LLERON
10-^
-
9lU
^ o
c
$
87- r, ,
6^
!
5V
-
4-J
3-
— _
21ft '
22
26
30
34
38
42
46
50
TIME (min)
Figure 27. Improvement in Variance from the use of PID Control.
54
74
The two histograms in
Figure 28 show the percentage of probes above a given
temperature. The histograms were constructed from averages of all of the temperatures
of thermocouple in the tumor during manual control (from T, to Tj in Figure 26) and
after (from Tj to T* in Figure 26) the controller was activated.
TEMPERATURE HISTOGRAM
100%-
Legend
90%-
Unifom Power
Before Control
80%-
After Control
70%-i
60%-i
50%-j
40%-i
30% i
20% J
10% i
0%^
36
38
40
42
44
46
48
50
TEMPERATURE (C)
Figure 28.
Temperature Histogram of all Controlled and Non-ControUec
Thermocouples in Tumor.
75
Figure 29 displays the maximum ARMA filtered temperature input to the
controller for each sector that contained thermocouples. The numbering and angular
arrangement of the sectors was shown in 3.
54-
COKTROUERON
53-
Lsgsnd
52-
o
UlX
is
ulS
a.(E
HI
— sectors
51 -
-- SdctorA
504948-
sectors
-10
- -
47-
sectors
— sectors
46- - SectorlO
45-
-- Sector 11
44-
43-.
4241-
5 rrmi^
4025
30
^ .
TIME(MC^
40
45
50
® ! 7\ ®
I
SECTOR NUMBSUNQ
Figure 29. Temperature Input as Seen by the Controller using ARMA Plus
Max Temperature.
76
3.5 PAIN BUTTON
For verification of the pain button system and accompanying software a series of
tests were performed on both the old algorithm described in Appendix B and the new
algorithm presented in section 2.5. The tests (listed Table 2) showed that the pain button
operated according to the algorithm described above and behaved as expected in the
various circimistances that occur during a patient treatment.
77
The start of a new scan resets all windows
Pain button software debouncing only accepts valid presses
Button press establishes a window behind the gantiy location at the time of the
press
Power in a window is at 0% during the first pass of the gantiy through the
window
Power in a window increases by 10% with each subsequent pass through the
window
Power for a window is not permitted to increase past 90%
Power in window is reduced by 10% with additional presses and held
Power in a window stays at its held value
Window follow all steps of the pain algorithm
Absolute ultrasound power changes when the gantiy enters a window and returns
when the gantry exits
Absolute power is a window is a fimction of sector power
Absolute ultrasound power changes while the gantiy is in a window and the gantry
enters a new sector of a different power setting
Gantry position is taken to be in the window of lower power if two or more
windows overlap
Windows stay in the correct position when screen scale is changed
Table 2. Pain Button Checks.
After the above verifications, the pain button with the old algorithm was clinically
available over a period of two years. Whenever a patient was to be treated with SFUS
the operator of the system would decide if this treatment had the potential of being
painful for the patient. The decision was based on the operator's past experience with
similar treatments or on the patient's own treatment history, if one existed. Deciding
factors included the tumors proximity to bones and the amount of medication
78
administered to the patient prior to treatment. From the operators standpoint, the button
was given to the patient if they were experiencing sharp pain that appeared quickly and
then faded in unison with the repetitive cycle of the gantry scan pattern. If the pain button
was not initially given to the patient, it would be made available if the patient e^qierienced
pain. The pain button was always given to the patient with the explanation that pressing
the button would allow a computer to help alleviate the pain. The patient was also told
that the button should be pressed as soon as sharp pains are felt.
3.6 PAIN BUTTON RESULTS
The pain system described in section 2.5 was an evolution of schemes. The
immediate predecessor to the one above is described in appendix B, and can be
summarized in three steps. A- Establish a window and look for the correct power. B If step one fails, move the location of the window back one radius. C- After three
executions of step B, the diameter of the window is enlarged up to a maximum of 150%
of its original size. The algorithm just outlined was available during the treatment of 28
patients over a two year period for a total 84 treatments. During these treatments, 65
complaints of unbearable / excessive temperatures or pain were made. Out of the 65
treatments, 26 of the patients experienced (and were able to identify) the sharp cyclical
pain that the pain button was designed to help alleviate. Data collected during the 26
treatments consists of 850 pain button presses which are broken down in the pie chart ia
Figure 30. Moving the location of the window occurred in 17 (65%) of the 26 treatments
79
RECORDED BUTTON PRESSES
(61%)
860 PRESSES
New Windows
LoworedPcww
Moved Wkxlcwr
IncwMed Radka
EndolAlooinhin
Figure 30. Breakdown of Recorded Button Presses.
in which the button was used. However, 95 (11 %) of the 850 recorded presses with the
old algorithm progressed of all of "the windows established progressed to the point where
the location of the window was moved.
A typical example of the pain button being used during a patient treatmoit is
shown in Figure 31. The pain button was pressed a total of seven times establishing three
separate windows. Before the pain butt(Hi presses, the patient complained of sharp pain.
All three of the windows progressed to step four of the algorithm explained in appendix
B whereupon the ultrasound power was adjusted to 90% in windows one and two and to
10% in window three.
After analysis of the data collected and observation of the treatments in which the
button was used, it was determined that the fifth step as described in appendix B created
a logic flaw in reasoning.
The flaw was that the old
Wlndowl
algorithm did not account
1
for delay times longer that
7
- 7
4C
2
the time required for the
gantry
to
travel
3 3 '
one
window diameter or pain
WMow2
15
.
Center
1
W)nda«3
,
Scanpatti
areas that were larger than
one window diameter. The
SCAN SIZE
flaw is seen when the
50x60
SCALE
30x40
lOmm
patient presses the pain
Figure 31. Patient Treatment with Old Pain Button.
button causing the pain
window to be relocated. The patients tolerance for pain is not be found because the
patient would eventually press the button near the first window location establishing a
new window versus adjusting the power in the first window in its relocated position as
needed by required by the patient. The first window in its new position would then
progress past the patient's tolerance causing presses in the second window. The second
window would also advance in the algorithm until it is relocated- starting the process
over. It was for this reason that the algorithm was deemed to have not been entirely
successful in effectively eliminating the patients pain for most of the treatments in which
then relocation step was reached.
Based on the above ratio, the logical flaw just
81
described, and the operators own experience, the new algorithm presented above was
designed and awaits clinical trials.
82
CHAPTER 4.0: DISCUSSION
4.1 TEMPERATURE CONTROLLER
The temperatures achieved with the addition of power control as a function of
position could still be improved, but appear to be more uniform when compared to not
having control as a function of position. In addition, the operator was not required to
monitor and adjust the power level. It can be seen in Figure 24 that increased damping
is required because of the oscillations in the temperatures seen during the operation of the
controller. This shows the need for the derivative implemented for the next series of
treatments and/or increased scanned scanning speed. In addition, initial treatments (like
the one shown in Figure 24) during the use of the PI controller and the first patient
treatment that used the PID controller (shown in Figure 25) contained dips in the
temperature when computer control was first activated. This demonstrated the necessity
for the bumpless transfer to automatic control that was outlined in section 2.4.2. The
histogram (Figure 28) shows that the increase in temperature uniformity was not at a
substantial cost in terms of the minimum temperature characterized by any area of the
histogram to the left of the 43 C° line. In retrospect, it is apparent from the histogram
of Figure 28 that the control temperature should have been set higher than 43 °C. While
it is not known what an ideal histogram would look like, an infinite slope at the target
temperature is a good acceptable goal. The increase in the slope of the histogram (caused
by activating the controller) also reflects the lowering of the variance seen in Figure 27.
83
Figure 28 shows that the temperature controller did what was expected- control the
power in such a way as to increase the uniformity of the temperature field. Figure 29
(the controller input with anna and max) shows that the temperatures settled into two
groups, those around 42.5 C° and those around 45 C°. It can be seen that the power
required to maintain the sectors at 42.5 C° prevented the temperature from decreasing
in the hotter sectors. The power level for the hotter sectors was set at 0% after the 45
minute maiic by the controller and the temperatures still did not fall. The power in the
colder sectors settled to 100% power at the same time. This shows that all temperatures
decreased when the controller was activated. It can be concluded from such variations
that thermal coupling, acoustic coupling, the effects of the boundary conditions, or the
effect of variations in the depth of the temperature probes existed between adjacent
sectors (because the controller saturated to it upper and lower limits in all sectors it was
controlling) these fectors and should not be ignored in future controller designs.
4.2 PAIN BUTTON
Ideally Figure 30 should show that 50% of the recorded presses established new
windows of the correct size to cover the area of pain and correctly accounted for the
patient delay time. The other 50% of the recorded presses should lower and hold the
power at the patients tolerance for pain. While a 50/50 split did not occur in the
recorded pain button presses, the pie chart of Figure 30 shows that the old pain system
recorded 61% of the presses as estabhshing new windows and 24% of the presses as
84
stopping the power from increasing and lowering the power in existing ones. This shows
that the button is useful during treatments.
Figure 31 shows that the patient was able to consistently identify the area of
discomfort by the use of the pain button. While the patient continued to indicate
discomfort, the relief provided by the pain button alleviated the need to move the scan
away from the center of the tumor volume.
Overall the pain button system has been shown to be beneficial during SFUS
treatments. The different types of pain sensations (skin and deep) that were discussed in
chapter I are no longer relevant to the operation of the treatment because the successful
identification of pain in this system is only dependant on the time at which the pain
occurs. For the same reason, the deep pain referral discussed in chapter 1 will not
adversely effect the pain identification because the computer knows the location of the
ultrasound gantry at the time of the pain button press.
85
CHAPTER 5.0: FUTURE WORK
There are many possible extensions to this work. This could include improvement
of the controller by further eliminating integrator windup by imposing a second restriction
on the evaluation of the integral of the error. This condition would only integrate the
error when the power level is not saturated at either of its limits. Such a condition would
prevent the area under the integrator term from growing in magnitude when no additional
changes in power could be taken in the direction required to zero the error (Franklin et
al. 1991).
Another way to avoid windup is to implement a velocity form of a PID
controller in which the proportional term depends on the derivative of the error instead
of the error itself (Astrom and Wittenmark 1984). To improve the temperature profile,
the controller could be given more resolution by dividing the scan area into more sectors
than presently used. A more sophisticated multi-input multi-output (MIMO) temperature
controller could be developed that could use fuzzy logic, or adaptive control laws to
replace the SISO PID controller. Also, an identification phase could be used at the start
of a treatment to find which thermocouples are coupled to each sector of the scan.
Another approach toward improvement of the treatment is to have the program
vary the frequency content of the ultrasound to determine which frequencies are most
effective in raising the temperature field and which frequencies are causing patient pain.
The use of multiple frequencies would allow adjustment while maintaining the total
absorbed power at a constant.
86
The program could also be made more intuitive by an improved user interface
other than the keyboard. This new user interface could include a mouse, light pen, or
touch screen.
87
APPENDIX A.O: PROGRAM USERS MANUAL
A.l APPLE COMPUTER PROGRAMS
The pain program communicates with two other computer programs. The first
is the scanning program on the Apple computer. The Apple computer should run the
program "TALK" which is identical to the program "XY0CTAG0N5P0TS" written by
Dennis Anhalt (University of Arizona. Cancer Center. Clinical Engineer) except that
TALK" sends scan parameters to this program over a custom computer bus. It does not
matter which computer program is started first as long as the pain program is started
before the first scan is initiated. If the Apple runs old scanning programs the pain
program will then not be able to identify any scan parameters or identify when the gantry
is in an inner or outer scan.
When the scanning Apple computer starts a new scan, the center of the gantry
display on the Vectra will automatically be updated similar to the MOVE CENTER
command except that when the center is set in this manner, the pain program goes into
an identification phase to search for scan parameters such as weather there is an inner
scan, the sizes of the scans, and the speed of the gantry. A message will be displayed
stating that this process can be aborted if there is no inner scan to identify by pressing
the "SPACE BAR". During this time, none of the menu items will operate until the
identification has been completed.
88
A.2 HP TEMPERATURE COMPUTER PROGRAMS
The second computer that the pain program commimicates with is the HP
temperature computer.
The program to run on the HP temperature computer is
#19:contlosc if the 1 MHz oscillator is being used exclusively or #19:cont2osc if the 1
MHZ oscillator and one Wavetek are being used. THE ORDER IN WHICH THESE
PROGRAMS ARE RUN IS IMPORTANT and running them in the wrong order will
result in the HP temperature computer locking up and requiring a complete restart by
pressing the SHIFT and RESET keys simultaneously. The pain program on the vectra
MUST be run before the "Y" conformation is entered to the prompt "Is all this data
entered correctly?" on the HP temperature computer and must continue to run as long as
temperatures are being taken. See the section on ENTER TUMOR TRACE AND TC'S
for more information on how th^ two computers communicate and for an important
warning. Other programs may be run on the HP temperature computer but the pain
program will not receive any temperature information preventing the controller from
having any effect on the treatment and preventing any temperatures from being displayed.
TO START THE PAIN PROGRAM TYPE "STP" FROM ANY DIRECTORY
ON THE HEWLETT PACKARD VECTRA QS/20 386 IBM COMPATIBLE
COMPUTER. STP is a batch file that runs the program PAIN1.EXE in the directory
BC.
89
THE PAIN PROGRAM WILL THEN PROMPT FOR THE PATIENT'S NAME.
If ENTER is pressed without entering a name, the default name of TEST will be used.
The patient name is used for the title of all data files. If a name is entered that has been
used previously, the program will alert the user and prompt for a unique name.
A.3.0 PROGRAM MENU
There are two methods for selecting menu items: The first way is to press the
"hot key" highlighted in red for each of the menu items. The pain program is not case
specific, i.e. "Y" is die same as "y". The upper left portion of the screen will then
prompt for the correct user input for that particular item. The second method for selecting
a menu item is to use the cursor arrow keys to move the menu pointer (V in purple) to
the desired item and press ENTER. Again, the upper left portion of the screen will
prompt for the correct user input."
The menu items are listed and described here in order of appearance.
A.3.1 QUIT: hotkey = "Q".
Used to end the pain program. When selected, the pain program will prompt for a yesno confirmation.
Only a correct response of "Y" or "y" will terminate program
execution at which time the operator is required to insert a disk to backup the data files
created during the just completed treatment. The program asks which of the floppy
drives (a; which is a 5 1/4 inch drive or b; which is a 3 1/2 inch drive) contains the
90
backup disk and then checks to make sure that the name is unique. If it is not the
program requires that a different disk be used without files of the same patient name.
A.3.2 WINDOW DIAMETER 20MM: hot key = "D".
Used to change to diameter of the pain button windows. This selection does not effect
windows that have already been established. Once selected, the " + " and
keys are
used to increase or decrease the diameter in 2 mm increments. The default diameter of
20mm is shown here. The menu list will display the current diameter. The acceptable
range for the diameter is from 10 mm to 98 mm. If either limit is exceeded the diameter
will rollover to the other limit.
A.3.3 CONTROLLER TARGETS; hot key = "T".
If the temperature controller is ON. the " + " and
keys are used to increase or
decrease all of the controller target temperatures for each sector by 0.1 C increments.
Individual sector target temperatures are independent of one another and can be changed
by moving the menu pointer to the desired sector and using the " +" and
keys in the
same way as described above. The valid range for controller targets is from 20 C to 50
C. If either limit is exceed the target will roll over to the other limit. The initial default
target temperature is 44.0 C.
A.3.4 ZOOM: hot key = "Z".
Used to change the scale of the gantry position display. The " +" and
are used to add
or subtract 10 mm from the scale of the gantry position display screen. The scale can
been read by the light blue grid on the screen which shows a 10 mm grid when there
have been no temperature positions and tumor trace entered. When temperatures have
been entered, the left and bottom edges of the display will show tick marks at 10 mm
increments. The minimum scale is 20 mm or the size of the inner scan (if there one
currendy exists). The maximum scale is 300 mm. The scale will be automatically
adjusted to a best fit after temperature positions and tumor trace have been entered. If
the screen is zoomed in smaller than the best fit the thermocouple and sections of tumor
trace that do not fit on the new screen scale will not be displayed. Also, if the gantry
bumps an edge of the display area the screen will automatically zoom out 10 mm to
accommodate.
A.3.5 CONTROLLER IS OFF: hot key = "C".
Indicates the current state of the temperature controller. This item is used to turn the
temperature controller ON or OFF. Pressing "C or ENTER when the menu selector is
on this selection will toggle the controller on or off. The menu will then display will
then toggle to CONTROLLER IS ON (or OFF). Each sector and its target temperature
are independent from one another. The controller will turn ON if there have not been
thennocouple locations entered with the ENTER TUMOR TRACE AND TC'S command
but it will have no effect. When activated, the manual power control will no longer be
92
available as will be seen from the power percentages for each sector changing to a target
temperature.
Once on, every time the HP temperature computer takes a reading, the
temperature in each sector that the controller is using for input will be circled in red and
the percentage power for that sector will be updated. If a sector does not have any
thermocouple in it, no power change will be made for that sector. IT SHOULD BE
NOTED that if the screen is zoomed in with the ZOOM command to a scale smaller that
the best fit scale that the pain program selected, the thermocouple at the edge of the
screen which no longer fit with the new scale will not be displayed but will still be used
by the controller unless they are deactivated with the VALIDATE command described
below. If the controller tries to use a thermocouple for input to that is not displayed
because of the zoom setting, the program will zoom out until the thermocouple is
displayed. When the controller is turned off, manual control will be restored. The
percentages displayed after the transfer from automatic to manual control will reflect the
last power percentages that the controller set. This was done to try and achieve a
"bumpless temperature transfer".
The controller was designed for a sample rate of 30
seconds. A sample rate different than 30 seconds will result in undershoot or overshoot
of the temperature profile. Therefore, THE HP TEMPERATURE COMPUTER MUST
BE SET FOR A SAMPLE TIME OF 30 SECONDS IF THE CONTROLLER IS GOING
TO BE ACTIVATED. The control law used is describe in detail in section 2.0.
93
A.3.6 REDRAW SCREEN: hot key = "R".
Used redraw the gantry position portion of the screen in the event of any "garbage
collection". This item has no other effect on the execution of the pain program.
A.3.7 RESET WINDOWS: hot key = "W".
Used to erase all pain windows that have been established with the pain button. This
option does not reset the count on the total number of button presses available. Requires
a "Y" or "y" confirmation.
A.3.8 MOVE CENTER: hot key = "M".
Used to change the center of the gantry position display screen to the current gantry
position. Requires a "Y" or "y" confirmation. The pain program's display of gantry
position can be visualized as a sli(fing view area over the range of gantry motion with the
center of this window defined by this command. The gantry position display will center
on the gantry's location at the start of the pain programs execution by default.
If this
item is used after the mmor trace and probe locations have been entered with the ENTER
TUMOR TRACE AND TC'S command, a small red circle will be displayed to indicate
where the tumor trace center is now located relative to the new gantry center.
This
command is executed automatically just before the apple computer starts a new scan.
94
A.3.9 ALIGN TUMOR TRACE: hot key = "A".
This selection is used to align the tumor trace and probe locations to the current gantry
location. Requires a "Y" or "y" confirmation.
BEFORE EXECUTING THIS ITEM
READ THE WARNING ON THE ENTER TUMOR TRACE AND TC'S COMMAND.
The only time this item should be used manually is when the alignment of the gantry
center to the trace center is determined to be incorrect, i.e. if the patient moves or the
apple was not started with the scan center aligned correctly to the tumor center. This
command is automatically executed when the first center is established either by the
MOVE CENTER command or by the first scan initiated by the scanning apple. This
selection also executes a MOVE CENTER command described above.
A.3.10 ENTER TUMOR TRACE AND TC'S: hot key = "E".
When the this command is select^, a tablet self test is performed. If the self test fails,
the command is aborted. The tablet is then configured in a remote request mode to allow
the pain program to acquire data at its own rate. If the default configmtion of a stream
mode is used, the data quickly overflows the serial buffer on the Vectra computer and
causes a program crash.
This selection is used to enter a tumor trace, thermocouple probe locations, a
center to align them to the gantry center, and an edge point to find the correct
orientation. When selected, the pain program will first check that the graphics tablet has
been turned on. If it has not, a message will be displayed asking that it be turned on.
95
In the rare event that the pointer on the screen is stuck in a comer, regardless of the
movement of the stylus, the graphics tablet did not initialize correctly. The remedy is
to reboot this computer by pressing CRTL-ALT-DEL simultaneously and restarting the
pain program. All tablet entries are on a 1 = 1 scale: one millimeter on the table equals
one millimeter on the screen. This allows the transparency being traced to be held up
over the computer screen to check for correctness. After entry the scale that the eateries
are displayed is dictated by the screen scale described in the section on the ZOOM
command. The orientation of the screen is a standard cartesian coordinate system with
the origin on the lower left portion of the screen. Y increases up and X increases to the
right. The graphics tablet is configured with the same coordinate system as the screen.
The first entry is the tumor trace. If a mistake is made and the user wishes to
start entering the tumor trace over again, press the "SPACE BAR". At this stage in
development, the tumor trace is only used for display and has no effect or meaning to the
pain program except for display purposes. There are four colored buttons on the graphic
tablet stylus that all act the same. The buttons are toggled with the upper left portion of
the screen displaying the current state of the buttons as either TRACE OFF or TRACE
ON. Thus, momentarily pressing one of the four buttons will change the state from ON
to OFF or visa versa. The tumor trace section has two data filters to limit the amount
of data input. The first filter requires that there must be at least 5 nun between data
points. The second filter demands that the angle between consecutive data points must
be less than 180 degrees. If the angle is more than 180 degrees, the computer will stop
96
accepting new points until the second condition is again satisfied. When the end point
of the trace reaches within 10 mm of the start of the trace, the pain program will
automatically connect the points and move onto the next stage.
The pain program will then prompt for the entry of the tumor trace center which
will be used to align the tamor trace and thermocouple locations to the gantry scan center
(see the warning below). Move the stylus to the desired center point and press one of
the stylus buttons. Next, a point somewhere on the edge of the trace must be entered
in the same manner. The edge point is used to rotate all information entered with the
tablet to the same angle as the patient. The pain program then asks for a confirmation.
Answering anything other than "Y" or "y" will allow the re-entry of the tumor trace,
center, and edge point.
The paiQ program will now want the number of probes the user wishes to enter.
The acceptable range is from 0 to 8. If the HP temperature computer program has been
executed past the "Y" confirmation at the "Is all the data correct?" prompt, the pain
program will have the correct number of probes to enter as the default. To except the
default number of probes just press ENTER. While the pain program will accept a
different number of probes, doing so will result in a "PROBE COUNT MISMATCH"
warning when the first set of temperature readings is taken unless the HP temperature
computer has been restarted with the new number of probes. If this mismatch warning
occurs, the pain program will show the number of probes it is receiving and erase all the
97
previously entered traces, probe locations, and tumor center so that new ones can be
entered. If the HP temperature computer has not run past this prompt the user will be
required to supply the correct number of probes. Again, if this number does not match
the numi}er of probes received during the first reading the action described above will be
taken and this section will have to be repeated. If the number of probes to be entered is
zero the pain program will ask for a "Y" or "y" confirmation before continuing. If the
last probe is a surface probe, respond with a "y" to the question "Ignore last probe".
This has the same effect as using the VALIDATE TC command for all seven
thermocouple of the last probe entered. Next, THE PROBES MUST BE ENTERED IN
THE SAME ORDER AS THEY WERE (ARE TO BE) ENTERED IN THE HP
TEMPERATURE COMPUTER. The tip of probe one is entered first- move the stylus
to the desired location and depress a stylus button. Next, enter any point along the probe
in the same manner as above that is at least 10 mm away from the tip point just entered.
The pain program assumes that the probes are positioned in the X-Y plane and are 60 mm
in length with seven sensors spaced 10 mm apart. Continue this process until all of the
probes are entered. Pressing the SPACE BAR will start the probe entry process over.
After the last probe is entered answering anything other than "Y" or "y" to the prompt
will also start the process over.
WARNING: THE TUMOR CENTER POINT MUST NOW BE ALIGNED TO THE
CORRESPONDING LOCATION ON THE PATIENT!!!
After the location entry
described above, the pain program will align the tumor/probe center with the current
98
gantry display center. If the gantry display center is not currentiy located at the center
entered on the trace entry, the display would be spatially incorrect and misleading.
Therefore, the user must first move the gantry to the edge point that was entered above
and use the ENTER EDGE POINT menu option described below. Then when the scan
starts, the program will determine the scan center which must align with the center
entered above. The trace and thermocouple will not be displayed imtil both the edge
point and center point are entered from the gantry position. When the first new center
is established by either
the MOVE CENTER command or automatically by the apple, the
pain program will execute a ALIGN TUMOR TRACE command to align the centers.
Therefore, the first scan of the gantry should be centered on the location on the patient
indicated by the center entered in this section.
A.3.11 SCREEN TARGET 42.5 C: hotkey = "S".
This command is used to change the temperature at which the thermocouple temperatures
displayed on the screen change from blue to red. A temperature displayed in blue
indicates that the thermocouple is below the screen target and a red temperature means
that the thermocouple is above the screen target. The screen target is completely separate
from the temperature controller target. The screen target is only used for an informative
display.
To change the screen target up or down the " + " or
keys are used
respectively. The screen target increments are 0.1 °C. The valid range is from 20 °C
to 50 °C. If either limit is exceeded, the screen target will rollover to the other limit.
99
The default screen temperature is shown here as 42.5° C and will always reflect the
current setting.
A.3.12 T/C VALIDATION; hot key = "V".
This command is used to prevent a thermocouple from being displayed and from being
considered for use as input to the temperature controller. The pain program will prompt
for the thermocouple to be ignored. The numbering scheme for the thermocouple are the
same as for the HP temperature computer.
Enter the two digit number or if the
thermocouple number is less than 10 and only there is only one digit, enter the single
digit and press ENTER. The pain program will then ask "IGNORE T/C X (Y/N)"
(where X is the thermocouple number being considered). Responding with " Y" will have
the effect of disabling the thermocouple. A "N" response will reverse a previous ignore
command. This command should be used when a thermocouple is known to be broken,
when a thermocouple is not in tumor, or the user just does not want a particular
thermocouple to be used. When a thermocoitple is ignored, the change is not seen on the
screen display until the next temperature reading or until the screen is redrawn. Only the
decimal points of a deactivated thermocouple are displayed.
100
A.3.13 ENTER EDGE POINT: hot key = "P"
This selection must be used after the traces are entered with the graphics tablet and before
the scan starts. The gantry must also be aligned to the location that was entered with the
ENTER TUMOR TRACE AN T/C option before selecting this item. Once selected the
program will ask for confirmation to the location entered. The edge point can also be
re-entered if the patient rotates his/her position. Patient movements in the X and Y
direction should be corrected by the use of the move center command witch is done using
the ALIGN TRACE command.
101
APPENDIX B: ORIGINAL PAIN ALGORITHM
The following algorithm describes the set of events for multiple pain button
presses in the same approximate location or window. Two pain button presses are
considered to be in the same window if the distance between them is less than the current
window diameter. The program is capable of handling up to 300 (an arbitraiy limit)
presses and multiple independent windows. For excessive numbers of windows the ^)eed
of the program becomes hindered as more and more regions are established.
The action of the computer from pain button presses and the reasons behind those
actions can be outlined in seven steps below and can be seen in Figure B1.
The overall goal of the algorithm used was to deliver as much power as possible
to the patient without causing excessive pain. To this end, the patient tolerance for pain
must be found. This is accomplished by the following set of predetermined actions to
possible situations.
STEP ONE:
Upon the first press of the pain button a circle of operator selected diameter (from
a default setting of 10 mm to a maximum of 98 mm ) is established around the
location of that the gantry was at the time the press (Figure 82V The ultrasound
power to the transducers is immediately reduced to 0% while the gantry is inside of
this window.
102
NO
posnx)N
cc
2
Btttthtmd
PRESS?
YSS
NO •
fl
,
a
9
•X
posmoN
tc
POMT
B , R
—•-©
posmoN
to
•»
by 10%
YES
fl fl
^
100«
s
posmoN
i
by 10%
HO
PRESS?
YES
POWBt-O?/
NO
VIM
Figure Bl. Old Algorithm for Pain Button.
YES
posmoN
Rby10%
upB>150%
103
STEP TWO:
WINDOW 1
During
each
subsequent
pass
through the window the power will
increase by 10%. This is to see if
the patient can tolerate more power
(the goal is to deliver as much
power as possible).
SCAN PATH
Figure B2. First Button Press.
STEP THREE:
If the power reaches 100% without any additional pain button presses, the window
will be removed. The reason behind this action is that the patient was experiencing
a one time pain. Or, thermal build up in the tissue or bone has occurred which
should have been relieved by the temporary reduction in power to that area. Or, the
press was an accident and should be ignored as it did not represent the patient's true
condition. (A late modification to this step has been made so that the power inside
of a window will not inaease past 90%. Also, the windows are no longer removed)
104
STEP FOUR:
Additional presses inside the window will cause the power to be reduced by 10%
from its previous value and be maintained at that value. This press is taken to
indicate that the power was increased past the patients tolerance for pain in the
window and the power should therefore be reduced. No additional power
modifications will occur with each subsequent pass through the window as long as
no additional pain button presses take place. If an additional press occurs while the
power is being held at a fixed value, the program will lower the power by another
10% and hold it for future passes of the gantry through the window. This step will
repeat for each press in the window until step five is reached.
STEP FIVE:
If a press occurs while the power is at 0%, The window is relocated back by one
radius (Figure B3).
This indicates
that the patients reaction time was
WINDOW 1
OLD LOCATION
slow during the first press and the
window needs to be moved back to
try and cover the area that is
NEW LOCATION
actually experiencing the pain. The
pain program then restarts the status
of the window at step 2 in the hope
SCAN PATH
Figure B3. Relocated Window.
that the actual location is now known and the power tolerance can be found.
STEP SIX:
105
Step five
is limited to three
repetitions per window. After
WINDOW 1
three, a button press inside a
window with zero power will
cause the window diameter to be
increased by 2 mm in a last
/
SCAN PATH
attempt to relieve the pain (Figure Figure B4. Enlarged Window.
B4). The computer then restarts
the window at step 2
STEP SEVEN:
Step six is only allowed to increase the window size to 150% of its original size,
after which no fluther action is taken.
APPENDIX C: WIRING CONNECTIONS
VECTRA QS/20
2801 50 PIN
IDC
1 a/d input chO
GANTRY
POTENIOMETERS
25 PIN DB
FEMALE
SIGNAL NAME
1
25
GND
25
GND
25
GND
3 a/d input chl
5 a/d input ch2
7 a/d input ch3
8
ROTATE
25
9 a/d input ch4
GND
TILT A/D
25
GND
Table CI. Connections Between the Gantry and Vectra.
10
VECTRA QS/20
2801 50 PIN
IDC
DUTY CYCLE
BOX 15 PIN DB
FEMALE
38 bit 1/0
SIGNAL NAME
Button press
29 bit 0/1
10
reset bit
30 bit 0/2
11
safety
27
15
D-GND
Table C2. Connections Between the Vectra and Duty Cycle Box.
VECTRA QS/20
2801 50 PIN
IDC
APPLE A/D CARD
WIRE NUMBER
35 bit 0/6
3 ch2 yellow
SIGNAL NAME
center indicator
36 bit 0/7
1 chO black
inner/outer indicator
' 'able C3. Connections Between the Vectra and Apple.
108
iC;
;'c
ail
^
t>Cc
^
6^
^
w^
>
At9'BTO"
1^
I
-L ...
i c«
I
^ orrr:
—
"8
—^
*«wao«
± ...
1^
—
d>-
1
. 5f
4>r-
rt>^
X-'
U--
It
Tv :
1
rXDE
Figure CI. Schematic of Dutycycle Controller Box
^X4r X *'
-I—r
5
109
APPENDIX D.O: GRAPHICS TABLET AND BOX CALIBRATION
D.l ACCURACY OF THE TABLET
It was necessary to determine the
GRAPHICS TABLET
accuracy with which data could be
JACCEPTABLE ENTRy AREA
2
2
entered by the use of the tablet. To
3
this end a short accuracy study was
undertaken. The result of the study
4.
4
*
3
1
showed that data points could be
entered within 0.2 mm (on average)
3
4
3
2
2
1
Figure Dl. Test Pattern.
with no edge effects or nonlinerarities.
This result was obtained by entering the end points of sixteen radial line segments divided
into four categories that covered the acceptable area for data entry for the pain program.
D1 shows the orientation of the segments entered and how they were categorized. The
entry of the segments was repeated five times with the results shown along with the
standard deviation for each category in table 1.
Line
Number
Line
Length
Standard
Deviation
1
21.01 mm
0.24
2
21.28 mm
0.23
3
21.00 mm
0.16
4
21.18 nmi
0.20
Table Dl. Graphics Tablet Test Results.
110
The true length of the lines was 21.2 mm. Since no trend can be seen in Table 1, it was
concluded that no edge effects or nonlinearalities existed in the data. These results are
consistent with the specification of the tablet which rate its accuracy at a maximum
resolution of 40 lines per millimeter (0.025 mm as the smallest distinguishable stylus
movement). The difference between the measured and specificated accuracy is attributed
to operator entry error.
D.2 DUTY CYCLE BOX CALIBRATION
It is necessary to calibrate the timing constant of the mono shot timer so that the
length of pulses are long enough to reach between the trigger pulses. The procedure to
calibrate one of the channels on the duty cycle controller is as follows:
1. Connect the input to the channel being calibrated to TTL high.
2. Connect the ou^ut of that chaimel to an oscilloscope.
3. Set the output from the Data Translation 2817 D/A card to its lowest setting by
ousting a digital value of zero using the PCLAB software.
4. Adjust the potentiometer located next to mono shot timer corresponding to the
channel being calibrated until the ou^ut on the scope shows close to a 100% duty cycle.
It is very important that the duration of the output from the mono shot timer not be
allowed to exceed the time between the trigger pulses.
unpredictable duty cycles.
Doing so would yield
Ill
After this adjustment, it is necessary to correlate the digital values to be sent to the
Data Translation Card that are required to produce each desired duty cycle increment
(0%, 1% 2% ... 99% 100%). This is done by connecting the channel to be adjusted in
the same manner as specified above except the output needs to be connected to a meter
that measures duty cycle (the screen of an oscilloscope can be used but the duty cycle
measurements are not as accurate). Once the values are found for each channel, a
polynomial needs to be found to allow the program to accurately interpolate between the
data points (third order is sufficient). The coefficients for the polynomial are entered in
the POSITION subroutine.
To change the frequency that the duty cycle controller operates, adjust the timing
constant of the timer with the potentiometer located next to the free running 555 timer
or by changing the capacitors located in the same area for different capacitor values to
achieve the desired frequency (see the Figure 15 for wiring details). The specification
sheet for the 555 timer contains a table that lists the required resistances and capacitances
for assorted frequencies.
After a frequency adjustment, it is mandatory that the above
calibration procedure be followed for all channels to assure that the correct duty cycles
are specified by the pain program.
112
REFERENCES
Anhalt D., "Two Methods for Relieving Patient Pain During Scanned Focused Ultrasound
Hyperthermia," Unpublished p^per. University of Arizona, Cancer Center, Clinical
Engineer, 1991.
Astrom K.J., and Wittenmark B., Computer ControUed Svstems. Theory and Design.
Prentice-Hall, Inc., Englewood Cliffs, New Jersey, pp. 180-1%, 1984.
Babbs C. F., Vaguine V. A., and Jones J. T., "A Predictive-Adaptive, Multipoint
Feedback Controller for Local Heat Therapy of Solid Tumors," A Transactions on
Microwave Theory and Techniques. Vol MTT-34, No. 5, pp. 604-611, 1986.
Buettner K., J. Applied Physiol. Vol. 3, p. 691, 1951
Clark W. C., "Pain Sensitivity and the Report of Pain: An Introduction to Sensory
Decision Theory," Anesthesiology. Vol. 40, Num. 3, Mar. 1974.
Coley W.B., "The Treatment of malignancy tumors by repeated inoculations of
erysipelas: with a report of 10 original cases. Am J Med Sci. Vol 112, pp. 251-281, Sep
1896.
Collins W.F., Nulsen F.E., and Shealy C.N., "Eletrophysiological Studies of Peripheral
and Central Pathways Conducting Pain," Pain, ed. by Robert S. Knighton and Paul R.
Dumke, Little, Brown and Company, Boston Massachusetts, pp. 33-45, 1964.
Connor W. G., Gemer E. W., Miller R.C., and Boone M.L.M., "Prospects for
Hyperthermia in Human Cancer Therapy, Part 11: Implications of Biological and Physical
Data for Applications of Hyperthermia to Man," Radiology. Vol 123, pp. 497-503,1977.
Corry P.M., Spanos W. J., Tilchen E.J., Barlogie B., Barkley H. T., and Armour E.
P., "Combined Ultrasound and Radiation Therapy Treatment of Human Superficial
Tumors," Radiology. Vol 145, pp. 165-169, 1982.
Corry P.M., Barlogie B., Tilchen E.J., and Armour E.P., "Ultrasound-Induced
Hyperthermia for the treatment of Human Superficial Tumors," Journal of Radiation
Oncologv Biologv and Phvsics. Vol 8, pp. 1225-1229, 1982.
Dewhirst M. W., and Sim D.A., "The Utility of Thermal Dose as a Predictor of Tumor
and Normal Responses to Combined Radiation and Hyperthermia," Cancer Research
fSuppl.). Vol 44, pp. 4772s-4780s, 1984.
113
EHederich C., Hynynen K., "Induction of hyperthermia using an intracavitary multi­
element ultrasonic applicator," IEEE Tran Biomedical Engineering. Vol. 36, pp. 432438.
Franklin G.F., Powell J.D., and Emami-Naieni A., Feedback Control nf Hynamir
Systems, Addison-Wesley Publishing Company, Reading Massachusetts, pp. 113-127,
1991.
Gemer E.W., Connoer W.G., Boone M.L.M., Doss J., Mayer E.G., and Miller R.C.,
"The Potential of Localized Heating as an Adjunct to Radiation Therapy," Radiology. Vol
116, pp. 433-439, 1975.
Harari P.M., Hynynen K., Roemer R.B., Anhalt, D., Shimm, D., SteaB., andCassady,
J. R., The University of Arizona Development of SFUS Hyperthermia: Clinical Respoase
Evaluation. Int J Radiation Oncology Biol phys. Vol 21 , pp. 1-10, 1991.
Hardy J.D., "Thresholds of Pain and Reflex Contraction as Related To Noxious
Stimulation," Jnnmal of Applied Physiology. Vol. 5, Num. 12, pp. 725-739, June 1953.
Hardy J. D., and Stolwijk J. A., "Tissue Temperature and Thermal Pain," in TouchHeat and Pain, ed. by A. V. S. DeReuck and Julie Knight, Boston, Litde, Brown and
Company, pp. 27-56, 1966.
Hardy J.D., Wolff H.G., and Goodell H., Pain Sensation and Reactions. New Yoric,
HafQer Publishing Company, 1967. pp. 86-122, 325-379.
Herman T. S., Teicher B. A., Jochelson M., Clark J., Svensson G., and Coleman C. N.,
"Rati(Hiale for use of Local Hyperthermia with Radiation Therapy and Selected
Anticancer Drugs in Locally Advanced Human Malignancies," Int J Hyp^ermia. 1988,
Vol 4, No 2, pp. 143-158.
Hynynen K., and DeYoung, "Temperature elevation at muscle-bone interface during
scaimed, focussed ultrasound hyperthermia." Int J Hyperthermia. Vol 4, pp. 267-279,
1988.
Hynynen K, Shimm D., Anhalt B., Stea B., Sykes H., Cassady J. R., and Roemer R.B
"Temperature Distributions During Scann^, Focused Ultrasound Hyperthermia
Treatments," Int J Hyperthermia. 1990, Vol 6, No 5, pp. 891-908.
Hynynen K., Roemer R.B., Anhalt D., Johnson C., Xu Z. X., Seindell W., Cetas T.C.,
"A Scanned, focused, multiple transducer ultrasonic system for localized hyperthermia
treatments," Int J Hyperthermia. Vol. 3, pp. 21-35, 1987.
114
Hynynen K., "Methods of External Hyperthermic Heating," Clinical Thermclogy.
Subseries Thermotherapv. Vol. 81, pp. 61-115, 1990.
Hynynen K. and Lulu, "State of the Art in Medicine, Hyperthermia in Cancer
Treatment," Investigative Radiology. Vol 25, pp. 824-834, 1990.
Iggo A., Recent Advances on Pain. Pathophvsiolo?v and Clinical Aspects, ed. by John
J. Bonica, Paolo Procacci, aiKi Carlo A. Pagni, Charles C. Thomas Publisher, Springfield
Illinois, pp. 3-81, 1972.
Johnson R.L., Kress R.B., and Roemer R.B., and Hynynen K., "Multipoint Feedback
Control System for Scanned Focused Ultrasound Hyperthermia," Phys. Med. Bio.. Vol.
35, pp. 824-835, 1990.
Kiff H. J., Pain Relief. J. B. Loppincott Company Philadelphia, pp. 1-41, 1974.
Knudsen K., and Heinzl, "Two point control of temperature profiles in tissue," Int. J.
Hyperthermia Vol. 2 p.21 1986.
KnudsenK., andOvergaardJ., "Tdentification of thermal model for human tissue." IEEE
Trans. Biomed. Eng.. Vol. 33, P. 447,. 1986.
Lele P.P., "Hyperthermia by Ultrasound," Proceedings of the International Svmposium
on Cancer Therapv bv Hvpeithermia and Radiation, pp. 168-178, 1975.
Lele P.P., "Physical aspects and clinical studies with ultrasound hyperthermia," In:
Storm FC (ed) Hyperthermia in cancer therapy. Hall Medical, Boston, pp. 333-367.
Lewis T., Pain, l^e Macmillan Company, New York, pp. 1-57, 1942.
Lele P.P., "Effects of Focused Ultrasonic Radiation on Peripheral Nerve, with
Observations on Local Heating," Experimental Neurology. Vol 8, pp 47-83, 1%3.
Lin W.L., Roemer R.B., and Hynynen K., "Theoretical and Experimental Evaluation of
a Temperature Controller for Soumed Focused Ultrasound Hyperthermia," Med. Phys..
Jul/Aug, pp. 615-625, 1990.
Lin W.L., "Scan Parameter Optimization and a Temperature Controller for Scaimed
Focussed Ultrasound Hyperthermia: A Theoretical and Experimental Study," PHD
Desertion. Department of Aerospace and Mechanical Engineering, University of Arizona,
1990.
Livingston W.K., Pain Mechanisms. The Macmillan Company, New York, pp. 44-80.
1943.
115
Marmor J.B., Pounds D., Postic T.B., and Hahn G.M., "Treatment of Superficial
Human Neoplasms by Local Hyperthermia induced by Ultrasound," Cancer. Vol 43, pp.
188-197, 1979.
Middletoa R.H., and Goodwind G.C., Digital Control and Estimation. A Unified
Approach- Prentice Hall, Englewood Cliffs, New Jersey, pp. 466-481, 1990.
Miller R.C., Conner W.G, Heusinkveld R.S., and Boone M.L.M., "Proiq)ects for
Hyperdiermia in Human Cancer Therapy, Part I: Hyperthermic Effects in Man and
Spontaneous Animal Tumors," Radiology. Vol 123, pp. 489-495, 1977.
Miller M. R., "Pain: Morphologic Aspects," New Concepts in Pain and its Clinical
Management ed. by E. Leong Way, F. A. Davis Company, Philadelphia, 1967.
Moran L., "The Meaning and Measurement of Pain," in The Practitioner Handbooks.
Pain and its Problems, ed. by Heneage Ogilvie and William A. R. Thompson, The
Blakiston Company, Philadelphia Toronto, pp. 28-39, 1950.
Oleson J. R., "A Review of Magnetic Induction Methods for Hyperthermia Treatment
of Cancer," A Transactions on Biomedical Engineering. Vol /bme 31, No. 1, 1984.
Overgaard J., "The Current and Potential Role of Hyperthermia in Radiotherapy," Int J
Radiation Oncolo^ Boil. Phys. Vol 16, pp. 535-549, 1989.
Overgaard K., and Overgaard J., "Investigations on the Possibility of a Thermic Tumor
Tber^. I. Short Wave Treatment of a Transplanted Isologous Mouse Mammary
Carcinoma," Eur J Cancer. Vol 8, pp. 65-78, 1972.
Overgaard J., "Simultaneous and Sequential Hyperthermia and Radiation Treatment of
an Experimental Tumor and its Surrounding Normal Tissue in Vivo," % J Ka^H^tinn
Oncolo^ Boil. Phys Vol 6, pp. 1507-1517, 1980.
Roemer R. B., "Optimal power deposition io hyperthermia. I. The treatment goal: the
ideal tenq)erature distribution: die role of large blood vessels." Int J Hyperdiermia.
1991, Vol 7, No 2, pp. 317-341.
Roemer R.B., Hynynen K., Johnson C., and Kress R., "Feedback Control and
Optimization of Hyperdliermia Heating Patterns: Present Status and Future Needs,"
Proceedinas of the Eighth Annual Conference of the A/Engineering in Medicine and
Biology Society, ed. by George V. Kondraske and Charles J. Robinson, Vol 3, pp. 14961499, 1986.
116
Shimm D.S., Hynynen K., Anhalt D., Roemer R.B., and Cassady J.R., "Scanned
focussed ultrasound hyperthermia: Initial clinical results," Int J Radiation Oncology Biol
Ehys, Vol. 15, pp. 1203-1208, 1988.
Stauffer P.R., Cetas T.C., and Jones R.C., "Magnetic induction heatiog of ferromagn^c
implants for inducing localized hyperthermia in deep-seeded tumors," TF.FH Trans Biomed
Eag, Vol. 31, pp. 235-251, 1984.
Suit H.D, "Hyperthermic Effects on Animal Tissues," Radiology. Vol 123, pp. 483-487,
1977.
Trembly B. S., Wilson A. H., Sullivan M. J., Stein A. D., Wong T. Z., and Stiohbehn
J. W., "Control of the SAR pattern within a interstitial microwave array through variation
of antenna driving phase", A Transactions on Microwave Theory and Techniques, MTT34, pp. 568-571.
Webster's New Collegiate Dictionarv. G. & C. Merriam Company, Springfield,
Massachusetts, 1982.
White J. C., and Sweet W. H., Pf^in. Its
and Neurosurgical Control. Charles
C Thomas Publisher, Springfield Illinois, pp. 9-31, 1955.