IEEE TRANSACTIONS 01\ BIOMEDICAL EI\GINEERI\lG. VOL
Communications
37. NO
II. NOYE:vIBER 1990
_ _______________
Computer-Controlled Dynamic Phantom for Ultrasound Hyperthermia Studies JON ZAERR, ROBERT B. ROEMER, AND KULLERVO HYNYNEN
Abstract-A
It consists of four
in vitro
preserved canine kidneys perfused with an
ethanol preparation by
a sin gle
pump, plus four computer-con
trolled valves and four flow meters. The flow rate to each kidney is computer-controlled, giving flexibility in the types of spatial and tem poral flow
devices [9], for developing and testing inverse algorithms [10], and for reproducin g the large scale physiological hlood perfusion re sponses of organs [1 I]. Two general approaches are available for
developing such phantoms: constructing a mechanical system from perfusable matrices such as sponges, andlor systems of tubes [7],
[12]; or using a preserved organ for in vitro studies [9], [13], [14]. dynamic phantom sy.tem for use in evaluating hyper
thermia heating equipment has been designed, constructed, and tested,
80%
III 'j
patterns possible.
Examples are given for both step and ramp
changes in flow, and for a temperature dependent flow which has been used to simulate physiological responses tn elevated temper atures ,
The latter approach has the advantage of possessing a physiologi cally more realistic perfusion pattern-at least but
for the organ itself,
not necessarily for tumors.
This communication describes a dynamic phantom system using this latter approach, which has been developed [151 for system test ing as described above, rather than for accurately producing actual treatment temperatures. It is based on the process developed by Holmes et al. [13] to preserve in vitro can ine kidneys in alcohol for repeated use. The present study extends the previous applica tions of this approach
to include multiple kidneys, with the alcohol
perfusion to each individual kidney controlled by a computer. (The
INTRODUCTION
During hyperthermia treatments the tissue temperature achieved
flow pattern within each kidney can have considerable spatial vari ation, and these are not controllable in the present system.) These
depen ds on the magnitude and distribution of the deposited power, thermal conduction, and blood flow. In order to evaluate thc clin
added features give flexibility
ical heating characteristics of hyperthermia systems, all of the above
both the spatial and temporal distribution of the flow rate.
factors have
uating the power deposition distributions of electromagnetic appli cators [I]. [2]. Similarly, water baths and static phantoms can pro useful
information
regardng
the
ultrasound
intensity
distributions needed for hyperthermia applications [3]-[6], These nonperfused phantoms have successfully provided absorbed power patterns for simpl e tissues. However. the use of "dynamic" phan
toms which attempt to simulate the blood perfusion characteristics of tissues has not been so successfuL The long range goal of dy namic phantoms is to combine the above static phantom data with simulated patient blood perfusion responses so that patients' tem perature distribution responses can be predicted prio r to a treatment for planning purposes. This goal has not been remotely approached due to the com plexity of reproducing both the tumor's and the normal tissues' blood perfusion responses. However. even though this long range goal remains remote, such dynamic phantoms can provide other useful functions
for hyperthermia in situations where approximate
tissue temperature profiles are adequate, and other system testing
and evaluation considerations provide the primary experimental goal.
For example, Baisch et at. [7] have used a simple set of parallel tubes embedded in a solid matrix to experimentally reproduce con ditions under which the bioheat transfer equation is valid. A second such goal
is
to pro vide a basis for evaluating feedback control sys
tems under conditions where blood flow can change. This method for experimentally
gives a testing the stability and robustness of
such controllers under approximate clinical conditions, An exam ple of such an application is that
Lin et ai, [8] who have used single
preserved in vitro canine kidneys for hyperthermia controller stud ies. Other possihle applications include providing a well instru mented experimental base for evaluating and optimizing heating Manusc ript received September I 1.1989; revised March 4.1990. This work was supported in part by NCI Grants CA 36428 and CA 33922. The authors are with the Department of Radiation Oncology and the Department of Aerospace and Mechanical Engineering. University � of Arizona, Tucson, AZ 85721.
IEEE
and simu
to be considered, Static phantoms have proven their MATERIALS AND METHODS
worth in providing a clinically relevant experimental base for eval
vide
in designing experiments
lating physiological responses since the user has some control of
Log Number 9038594.
Dynamic Phantom Description
Following is a summary of the system used in this study. The details of all of the system hardware, software, and calibration data are provided by Zaerr [15]. An Apple II P lus computer controlled the flow rates to, and recorded the temperatures in four kidneys mounted in a sonication chamber, as shown in Figs.
sonication chamber
I and 2. The is an 80% alcohol-filled Plexiglas container
with a thin mylar membrane on the bottom to transmit ultrasound. A variable speed gear pump (Micro Pump, In c , ) forced the fluid through the system, and the flow rates to each kidney were set by stepper motor-controlled valves and measured with rotating disk flow meters (Digital Precision Flowmeters, Inc.). The differential pressure across the gear pump head was measured with a pressure transducer (SenSym Model SCX 15DNC). The analog voltage out put of the transducer (0,25-9.25 V) was appl ied to one of the channels
of an analog-to-digital convener card on the Apple Com
puter. An external interface was designed and constructed which: provided power to and multiplexed the drivin g signals to the step per motor driver; conditioned the signal from the flow meters; and amplified
and multiplexed the temperature signals. Communication
between the computer and the external interface was provided through a 16-b parallel interface and a 16-channel, 12 b ADC. As
noted earlier, the kidneys were preserved using the alcohol
fixation process developed by Holmes et al. [13], The last stage of
rv ation process involved perfusing the organ with 80%
their prese
ethanol, which was the working fluid of the dynamic phantom pre sented here. Using this liquid minimized biological degradation of the phantom while maintaining ease of operation since the normally suggested rehydration
process was eliminated. Thc ultrasound ah low compared
sorption in the alcohol-lilled phantom is relatively
to other, normal tissues (e,g" muscle), as is the absorption in nor mal kidneys with their high water contents. We have
no reliable
quantitative data on comparative values. The flow control temperatures in the phantom were measured with 16. bare-wire, manganan-constantan welded junction ther mocouples (wire diameter 50 I'm) referenced to an AD590JH in tegrated circuit temperature sensor. These flow control thermocou-
0018-9294190/1100-1115$0 1.00
@ 1990 IEEE
1116
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 37. NO.
_______________
APPL�
£}!!!�!!.!�J:.t::.�"_!!'?¥__
II PLUS
COMPUTER
I.
NOVEMBER 1990
50.-------�------ --�
___________ _
FLOW RATES
" /
,
, .
·,1'
is X
Fig. 3.
,
,,/',1''''
,@ / ,"KDI NEYl
z 10 o i= U - 10 W a:: -30
DISPLAY OF FLOW RATES. TEMPERATIJRES
,
KIDNEY 3
30
USER INPUT
, ,
.'
-30 -10 10 V-DIRECTION
50
30 [MM]
Thermocouple and scan locations. Kidneys one and three con
tained thermocouple, as indicated by the X's. The ultrasound focus was scanned alternately on the two octagonal paths.
In each mode, an internal feedback control system ensure that the flow rate called for
by
produced. First, at flow rates above 35 Fig. I. Block diagram of the dyn amic phantom. The Apple computer con trolled the flow rates to the four kidneys in the sonication chamber and m easured the temperatures and flow rates via the external interface. The hyperthermia system measured additional temperatures and provided the heating power.
from
was
used to
the software was accurately
mL /min,
the output signals
the flow meters were used directly in a proportional control
algorithm to obtain the specified flow rate. In this mode, the valve position was varied to
force the measured flow rate to track the software. Second, at flow rates optical disks in the flow meters did nol spin
desired flow rate demanded by the
below 35 mL / min the
evenly and thus yielded erroneous measurements. Therefore, the valves were set based on the nearly linear dependence of flow
Flow
Stepper
Meters
Motors
rate
on the pressure rise and valve position, which was approximated
as: V = V,(O/O,) (P/P,.) where V = flow rate in mL / min, II = valve position (zero is closed), P = pressure, and c indicates a
Flow Control Valves
Thermocouples
4
Sonication Chamber
calibration point. Measurements from the /low meters at the lowest measurable rates were used for V", lip and Pc. The !low rate esti mates using this technique matched the actual !low rates to within 2 mL /min. Using these two valve positioning schemes, the largest expected error was 4 mL /min at a flow rate of 150 mL / min (the expected upper limit of measurement), the largest expected error
decreased to 1.2 mL / min at a flow rate of 5 mL /min [for tion details see Zaerr (15)]. Hype rt her mia
Differential
r---"'J Pressure Transducer
Thermocouples
calibra
Hyperthermia System To heat the dynamic phantom the hyperthennia treatment system described by Hynynen et al. [16] was used. The hyperthermia ther
Return line
mocouples were distributed in the phantom as shown in Fig. 3. The
Variable Speed
Gear Pump
kidney edges were located with ultrasonic B-scans from the hy
Fig. 2. Schematic of the dynamic phantom. 80% ethanol was pumped with a gear pump through four flow meters and valves into the four kidneys. The valves were operated by the computer controlled stepper motors. The computer also read the flow meters and the pressure rise across the pump.
perthermia system, and the thermocouples were located with the transducers as descrihed hy Hynynen et al. (16). the X's represent thermocouple locations as measured to within 2 mm . A focused ultrasound transducer (frequency 1 MHz, diameter
treatment
=
=
13 em, radius of curvature = 25 em, acoustic efficiency = 70%) was scanned in a translational manner through the phantom in two concetric octagons as shown in Fig. 3, with a total scan time
pies were connected to an analog multiiplexer in the external interface and amplified by a gain of 1000. This signal was directed
to the input of the ADC in the Apple computer. The thermocouples were calibrated to within ±0.25°C, In order to extensively monitor the temperature field in the phantom, additional multijunetion ther mocouples (outside diameter 0.7 mm fused silica tuhes), denoted as "hyperthermia thermocouples" in Fig.
I, were inserted into the
kidneys. These hyperthermia thermocouples were connected to an
ical power of 15 W was provided by the heating system. Finally, Hynynen et af. [17] have reported results with four overlapping
beams a t various perfusion levels,
flow rates in
Basic program was written to control the kidney either of two modes: programmed flow, in which a
showing that the heated region,
defined as th e area within which the
temperature
was 75% of the
peak temperature, extends about 5 mm beyond the perimeter of the scan path. Thus, the heated region was approximated as including a 5 mm margin around the outer octagonal scan path.
ex isting scanned, focused ultrasound hyperthermia heating system [16]. An Applesoft
of
9 s. In all tests described in this communication, a constant acoust
RESULTS-DISCUSSION Several test results are presented here to demonstrate the capa bility of this system:
I) programmed flow rates with constant power
list of flow rates and their time durations was preprogrammed, and
input (PF/CP); and 2)
the computer simply stepped through this list; and temperature con
stant power input (TCF/CP). In the first programmed flow test il
trolled flow, in which the total flow to each kidney increased or
lustrated, the temperature response of the phantom to step changes in perfusion was investigated. One su ch response is shown in Fig.
decreased as a programmable function of the kidney temperature as measured by the above flow control thermocouples.
temperature
controlled flow rates with con
4 for kidneys one and three. The th ermocouples shown represent
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 37. NO.
IT
:;... w c:: :::J
44 42
........ .. ._...... ....... _.. .... ...__ .
_
.
U
•
:;... W a: :J
� \J \
40:r-'\
!;( c::
38'
�
34
W Q. :::;:
15 MLIMIN...
..��"
H .r"."
1117
II. NOVEMBER 1990
�a:
W Q. :::!! W I-
36
-��---����32��o 5 10 15 20 25 30 35 40 45
44 42 40 38 36 34 32 0
10
5
IT :;... w a: :::J I< a: w Q.
:::;:
w I-
.. .. ......�. .�. .. .�.��_��_..__.. ... __..f�·�·". .. ·····r-- r�..... � �
44
_.
42
·· ·
Z
","., . ............. ,
:i :J �
40
w I< a:
38 36
::
!
34 32 0
5
10
15
20
15
20
25
15
20
25
(a)
(a)
25
30
20 18 16 14 12 10
8
0 .... u.
35
40
6
0
10
5
45
TIME [MINUTES]
TIME [MINUTES] (b) Fig. 4. The s im ulta neou s temperature responses to two kidneys to step chang es in p erfusio n under constant power heating. (a) Minimum and maximum temperatures in the heated region of kidney one, and the flow
Fig.
5.
(b) Temperature [e'ponse to a ramp change in perfusion. (a) Maxi
mum and minimum temperatures in the heated region of kidney three. (b) Flow into kidney three.
rate into kidney one. (b) Minimum and maximum temperatures in the heated region of kidne y three. and the flow rate into kidney three.
30
38
the upper and lower bounds on the temperatures measured inside the heated region. In region one. as the flow was increasell from IS to 25 mL/min, the temperature dropped about 2.5°C. When the flow rate was brought back to its original value the temperature increased to its previous level. Kidney three showed similar results for three-step changes in flow rate. These changes in temperature demonstrate the sensitivity of the phantom temperatures to changes in flow initiated by the computer, and they also show that perfusion changes in one region did not significantly affect the temperatures in the second region. To test repeatability. this series of flow rate changes was repeated three times with very similar results. The temperature changes associated with each change in flow rate for these three repetitions wcre compared, and the maximum differ ence between runs was less than 22% for all cases (based on the largest temperature change of the three runs). In a second programmed flow test the temperature response for a ramp change in How was measured (Fig. 5). As the flow rate increased, the temperature decreased. This illustrates the phan tom's versatility in providing various programmed temporal How patterns. (The temperature time lag in the hotter thermocouple's response is probably due to it's being located in a region of low perfusion, and is thus cooled mainly by thermal conduction.) The temperature dependent flow gives one the ability to study various algorithms (one can program the flow response as almost any function of temperature) which could possibly reproduce the temperature dependent perfusion observed in vivo (e.g., [II], [18], [19]). For example, we have used this apparatus [20] to system atically reproduce the oscillatory response of in vivo canine thighs to hyperthermia [II]. For this TCF ICP Study, the flow rate to each kidney was controlled with the algorithm: Vet) Vb + G*to(t td) where V flow rate in mL/min, Vb basal flow rate in mL/ min, G gain in mL/min per °C, to T temperature ele vation in °C, t time, and td response delay. Selected results from that study (to illustrate the use of this apparatus) are shown in Fig. 6 for a test with Vb 8 mL/min, G 4, and a time delay of 4 min. One can see that the highest flow ratc in kidney three is 28.4 mL/ min which occurred at 14.5 min. This is about 4 min after the peak temperature (38.8°C measured by the flow control
� w c:: :J I< II: w Q. :::;: w I-
25
:J
36
20
3:
.
""""""
10
0
(a)
10 0
.... U.
FLOW
20
30
B
5
30 25
38
Z :i :J
� 20 W
36
15
TEMP
34 .__•....•••.•.....•
32
W l-
a:
34
40
w a: :::J I< a: w Do :::;: w l-
�
15