Multiple sensor optical thermometry system for application in clinical hyperthermia

The thermometry system described is based upon the temperature dependence of the band edge absorption of infrared light in GaAs crystal. The design of the thermometry was completed, and the system was subjected to an extensive evaluation, including testing with tissue phantoms and microwave applicators. The system has up to 12 temperature sensors which are packaged in two basic probe configurations: a single-sensor probe with a length of 1.2 m and a diameter of 0.6 mm; and a four-sensor linear array probe with a length of 1.2 m, diameter of 1.1 mm, and spacing of 1.5 cm between adjacent sensors. Results of thermometry evaluation are presented, including data on automatic calibration, temperature accuracy and stability, and EMI protection.

168 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-31, NO.1, JANUARY 1984 Multiple Sensor Optical Thermometry System for Application in Clinical Hyperthermia VICTOR A. VAGUINE, MEMBER, IEEE, DOUGLAS A. CHRISTENSEN, MEMBER, IEEE, JOE H. LINDLEY, AND THOMAS E. WALSTON AbstracI-'-The thermometry system described is based upon the temperature dependence of the band edge absorption of infrared light in GaAs crystal. The design of the thermometry was completed, and the system was subjected to an extensive evaluation, including testing with tissue phantoms and microwave applicators. The system has up to 12 temperature sensors which are packaged in two basic probe configurations: a single~sensor probe with a length of 1.2 m and a diameter of 0.6" mm; and a four-sensor linear array probe with a length of 1.2 m, diameter of 1.1 mm, and spacing of 1.5 em between adjacent sensors. Results of thermometry evaluation are presented, including data on automatic calibration, temperature accuracy and stability, and EMI protection. INTRODUCTION THE laboratory and clinical investigation of hyperthermia has now progressed to a stage where the engineering com· ponents involved are critical to a careful evaluation of the technique. This includes the development of heat applicators (both electromagnetic and ultrasonic) which will produce rea­sonably well-defined and perhaps controllable heating patterns, and temperatu're monitoring systems which are accurate and convenient to use in the presence of the heat-inducing energy source [1]. This paper describes the development of a fiber optic temperature probe system intended for use during elec­tromagnetically induced hyperthermia. When electromagnetic energy is employed to heat the treated region, temperature measurement problems may be encoun­tered if traditional temperature probes, such as thermocouple wire p'airs or thermistors with leads, are immersed in the fields while the power is on. This is due to the finite conductivity of the wires leading to the sensor and, to a lesser degree due to its small size, the conductivity of the sensor itself. In order to eliminate or reduce interference errors, "non­perturbing" temperature probes have been designed. Develop­ment of these "nonperturbing" probes has followed two paths: 1) a thermistor as sensor in combination with high-resistance lead wires such as carbon-impregnated plastic, investigated by Bowman [2] and Larsen [3], and 2) optical fibers, either glass or plastic, as leads attached to an optical temperature sensor. A variety of optical sensors have been investigated, each tech­nique possessing its particular advantages: Cetas [4] has ex­plored the use of a birefringent crystal whose optical rotation Manuscript received March 1, 1983; revised July 15, 1983. V. A. Vaguine, 1. H. Lindley, and T. E. Walston are with CUni­Therm Corporation, Dallas, TX 75243. D. A. Christensen is with the University of Utah, Salt Lake City, UT 84112. changes with temperature; a liquid crystal sensor has been pur­sued by Rozzell [5] and Johnson [6]; and fluorescent sensors have been developed by Wickersheim [7] and by Samulski [8]. The fiber optic probe system described in this paper is based upon another type of optical sensor. It uses a small crystal of the semiconductor gallium arsenide (GaAs) whose optical absorption at a specifically chosen wavelength is sensitively related to the crystal's temperature [9]. The amount of optical signal returned after passage through the sensor may be de­tected and electronically translated, after calibration, into an indication of the probe's temperature. Details of the optical processes involved in the semiconductor sensor are discussed. SEMICONDUCTOR/FIBER OPTIC TEMPERATURE MEASUREMENT TECHNIQUE Semiconductor materials owe their unique conduction properties to the nature of the allowed electron energy states. What differentiates semiconductors from insulators and con­ductors is the forbidden energy gap Eg between the allowed energy bands. For semiconductors to be useful as an optical sensor, the energy gap Eg should satisfy the following con­dition: thus, very few electrons have enough thermal energy to be excited across the energy gap. When light is passed through the semiconductor material, photon energy may be sufficient to excite valence band elec­trons into the conduction band upon collision between the photons and the plentiful valence band electrons; thus the colliding photons are absorbed in the process. A sharp rise in photon absorption occurs when the photon energy exceeds the gap energy. This phenomenon is known as "band edge" absorp­tion in semiconductor. Experimentally, it has been found that the absorption coefficient follows an exponential curve near the threshold wavelength Ag = hc/Eg in accordance with Urbach's rule [10]. This variation in attenuation versus wavelength is shown in Fig. 1 where light transmission through a 250 11m thick sample of GaAs is plotted as a function of optical wave­length at two temperatures, 25°C and 40°C. The band edge is shifted toward longer wavelengths as temperature increases, yielding a negative temperature coefficient. This temperature behavior has been exploited for tem­perature sensor development based on the transmission of nar­row- band optical radiation with its central wavelength situated 0018-9294/84/0100-0168$01.00 © 1984 IEEE VAGUINE c •• '.: MULTIPLE SENSOR OPTICAL TH~Rr.tOr.tETRY SYSTEM ... .. I .. 1 o.~. a.2_ ,""'. I••• •• 1 ,, ! .. ,, , .." , ........ ' , -- , V ~ ,, , ,, • . N •• ." ." ." ".'-'" (j.oO) - l. Tho optkal II>JOrption odie of GIA. "' '10'0 difrCttnt ten>­pe .. tura ( ... lid Un<). Tho duhod lin< "'ow, tho .....,lJum of I Nllo .. -t>.nd JOU,,,,,. w!thin the stc(!p slope of the band edge. Fi&. I lh..,wl the spec· trum of such a 5<.Iurce superimpo5/:d on the ab5<.lrplion curve~. TEMP~RATURE SIi:NSOR CON~'IGURAnoN In our scheme. the optlcal radiation is transmitted to the $ensor by one of a pair of !./l1ll1l·diameter plastic fibers. The sensor il shaped In the form of a rmofleelOf prism whereby the radiation which pu.5/:S IhrOUgh the 5/:nsor is ooUe"ed by the other fiber of the pair. Fi&. 2(a) shows the conr'iut1ttion and dlmen~ions of the sensor design. This "r.ngement of separate transmit and receive fiber& considerably !impUfie~ the optlcal deugn of the SOurce and detector module since h elim· inates the requirement for bum splitters and lenses al the in· terface between Ihe fibe" lIld the SOUrCe Ind deteclor. ThCiC optical oomponenls ate of len bulky, lossy, and difficult to align. The conductivity of the semioonduclor 5/:nsor is less by many orders of magnitude than that of I comparable metallic pari. For .:<ample, resistivity of Ihe intrinsic CaAs is lOs n ' em. The ,,"aU sensor VQlume further reduces any effect of a mis­malch between tissue and PIObe conductivities. After attach· ment, the sensor and fibers are sheathed in Tenon tubinl whOit Inner diameter is 0.3 mm and OUler dilmetci is 0.6 mm [see Fill. 2(b)l; Ihe end of the tube is heat sealed wllh a liny Teflon plUj. Plulic fibers arc cho5/:n for their $ffiaU outer diameter (0.12 mm) 1Ild. their nexiblllty compand 10 glass fibe". The Illge numerica! apertule (NA = 0.58) reduces the effect of bending lossts, i.e., Ihe lOll of optical power in the fibe, ploduced by sharp bendinl of Ihe fiber [Ill. which appears as a change in temperalure to the clectronin mQdule. One disadvantage of plaslic fiber& is rclated to the relallvely high oplical loss at I center wavelenllh of 905 nm due 10 light absorption by a hydroxyl radical. Asa ruult, Ihe overalllcnllh of the probes Is limiled 10 apptoximately 1.2 m. The base of the probe consiS\! of an L-shaped metal hous· ing which contains Ihe LED source (light emllling diode) and the sllioon phOlodetc<: lor (PO) associated wilh each plObe. (0) Fia. 2. Cono,ulltlon of sinai'" and mullipk>t¢mpell""" ","obos. (.) The oomlccrductor .. ...,r and iu 1111<hmenl 10 lhe oplk fibeno (b) SiqI< oen",r r.mpe::o'u~ ","obe. «) Fo"'·~n .. r ,empell'u", Pf()\>e. FI;i. l. SI",," ........ probe (Iof!) and fou,. .. n .. , Uneu .... y probe. Both arc shown with their int .... tod modulo bueo. dolOd .nd oalh.., .... ood h~podnmic: ......t1es. Fi&. 3 shows a photograph of a slnsle sensor probe wilh Its ba5/: and a preampliflel PC boald. The oplical componenl! are epoxied dilcctly to Ihe fiber ends: this \nttglilion of Ihe fiber to the optical support devices substantially improves light transmission and declca5/:s the optical 10S51:s. Compiled 10 a previous design employing IWO metal fiber optic connectors per probe. this new integrated tech.nique yidd. approximately 6-8 limes mOle light ]>OWer transmitted to the photo-deteclor for • given length of probe, thus substantially improving signal·lo, noise IIlio, which ClIn be ".nslated 11lI.., improvemenl of Ihe,momeuy performanee and Increase of probe length. The l.ED is a moderate·pow., GaAs LED, chQSCn to provide an appropriale cemer wavelength to match Ihe absorption band edle at 0.905 101m. The mnsmit fibel is epoxied dilcctly to the emillin& surface of the LED aftel the prOtective !;Over has been removed: this maximlzes lIansmil1ed powel [121. The receive fibel is placed againsl the photod.elCctor window (as dO5/: IS po$Sible to the detection surface) and epoxied In place. 170 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-31, NO.1, JANUARY 1984 :rcnt pane} : cclc ~t8::-ceT: s· ... ·itc:-'es and LEi, diE plav5 ,. ~ ","sense!' fi1 E:, ? - a " ~ AI" ,;J .1-- ~ .• rLi ~ co ~ -- ro x I I ,ro I I 1TIiCTCpTOC€eSCr I I 5vEterr': I I • clock ,/ sensor t/12 I • EF"Cl<: .L. • FAI'~ .... --- --- .:8> ~ ~ .l, platinum · pa,allel resistor ar;c s€Tial V\N\J\JW\JV\ interfaces l' '" ,,. , cooler '" ae ~ ~ e· co r~o cali bra ti on well: "ro RS-232C " platinum resistors If r, ~ J, peripherals /\MIVVVV\ or external computer T TE heater/cooler Fig. 4. A block diagram of the thermometry system. The small total size of each sensor and its attached fibers has allowed us to develop a multiple·sensor probe, with several sensors arranged in a linear array inside a common sheathing tubing. Four·sensor arrays have been built with equal spacing of 1.0 or 1.5 cm between neighbor sensors. The four·sensor probe tubing has an inner diameter of 0.55 mm and an outer diameter of 1.1 mm [see Fig. 2(c)). The base of this multiple sensor probe is a combination of individual bases, and can be installed in place of individual probes in the electronic con­sole. The module base of each probe is firmly attached to a temperature-controlled plate inside the electronics console, called a "cold chamber." THERMOMETRY DESIGN AND FUNCTIONS Early design of a single sensor probe, based on GaAs/fiber optic technique, has been developed by D. A. Christensen [9) and used initially as a part of a 433 MHz microwave hyper· thermia system [13). Here, we describe the latest design of the GaAs/fiber optic thermometry system with multiple sensor capability. A block diagram of the system is shown in Fig. 4. The major components include sensor modules, a cold chamber, a calibration well, microprocessor·based electronics, and a front panel with switches and digital displays. The cold chamber is a thermally insulated housing with room for up to 12 sensor bases which are installed on a metal plate in the cold chamber. It was found that the central wavelength and the spectrum _ of the emitted light by the LED are affected by temperature variations of the LED base. Therefore, temperature stabiliza· tion of the metal plate was provided in order to assure a high degree of stability in the light spectrum and the central wave­length during system operation. The temperature stabilization is achieved by the use of the microprocessor-based electronics with a platinum resistor as the temperature reference. The microprocessor system is based on 8 bit Z80 CPU with a 12 Kbyte EPROM (electrically programmable read only memory) and an 8 Kbyte RAM (random access memory). Parallel interfaces are included for communication to other hardware in the system, and a serial RS·232C port handles all communications to external peripherals or computers. A sec­ond serial RS·232C port is available for future expansion. Thermometry system functions fall naturally into two categories: basic temperature measurements and the operator interface (front panel and peripherals). The microprocessor system controls both functions. Primary operator control of the thermometry system is through three dedicated switches and two digital displays on the front panel (Fig. 5). The power switch turns the sys· tern on and off. The calibration switch initiates an automatic calibration. By using the sensors select switch one can display temperature on the front panel for a selected sensor. The dis­plays are bright LED's specifying the sensor number and its temperature. The automatic calibration is accomplished by inserting all sensors into the calibration well and initiating the calibration VAGUINE ~'al.: MUI,TlPLE SENSOR OPTICAL THERMOMl"rRY SYSTEM '" FIJ. S. An CM:r:ill vl<:W of ,ht thcrmomtlIY .Y .. .." with 12 ~ .. ...." lempen.,u,e prob4 process. During this mode. Ihe calibration well Is Inhllllly cooled to less tNn WOC and Ihen warmed in 2_3" steps to approximately 55"C. A, each step, the calibration well 1$ held at a constant temptr.,ure Ions enough to a~cUJat.ly measure and store all sensor readings in RAM'I. including lemptrature dlla from th, •• high ple~ision pbtinum sen50rs used II tern· peratule references. The calibration process takes approxi. mately 9 min and provides da ll neceuary to mcaSUI. tern· perature ovel the ,ange of 20-5S"C. The calibration dlta table is used to oonvel1 sen50r readings (optical transmitted power) into temperamle readings. Durina ClIlibr.llion. the proarelS on callbrlllion is displayed on the fronl panel In lenns of re.1 time lemptralure of the calibra tion well. Pan of the RAM memory. where calibration data is stored. is powered by a bat· tery when the thermometry 1$ unplugged from an e~ternal power lin •. Although basic functions provided by the from panel oon· trol Ire sufficient to operate and monitor the system, pedph. erals. such as vide<.> display and printer. can be connected to RS·232C port in order to exp,and Ihe monitoring capabUilies into more "'phillicated leal·tlnte. visual. audio. and hard copy forms for all sen50rs. PERFORMANCE ANI) CLINICAL ApPUCATION Th. thermometry has b<:en tested and evaluated, including measurements with tissue pNntoms and m;Crow .. e applica· tou. Immediately after calibration, accuracy of , II p.ob<:s il within ::to.loC. Temperature ddfltesting hal indicated that to obtain ab50lute accuracy of ±O.:tc for all probes, calibra tion has to be performed wi thin the lin 8 h. The loIS of '~CUI'CY after 8 h. however. is minimal. and even I flel two days of optllIlion it app,oaehe. only iO.3°C. Throughout the testing. It wu ~ommon to observe several senson ,eadlng within one hundredth of one dC8lce of each other. The system is essentially Immune to EMI effect for the fre· quencie-s used in microwave hyperthermill. For eumple, using a well 1Il111ched 915 MHz microwave applicator aimed diJectiy u the thelmomelry system. no noticeable temperature leadilli effect was observed until microwave powel density, measu.ed at the thermomelry~ rose to _level of appro ximately S W/cml. which substanlially exceeds any expected microwave stray radiation during hyperthermia treatment. The flb<:r optic prob<:s exptlience no accuracy loss nor produce noticeable electromagnellc field perturbation In electromagnetic fields. Accuracy of the probe readings can be affected If the prob<:s are bent or pinched dulinS temptlilUre measurementl. Testing has lIiown that a b<:nd radius of I em degades accu· racy by 0.1". However. this effect was reversible on~e a prob<: was unbent. No appreciable effect wu obse rved for "dii glcalcl than 2.S cm. In clinical applications, Ills doubtful that b<:nd radll 1Illall enough to cause Ipplecitble error wm be re· qulred to place the probes a\ treatment sites. The b<:nd effect means. rather, that CIIe lIiould be taken to avoid pinching or kinking the probes. The mul tiple sensor (linear array) p.obe, lince It consists of four sensors bundled tOjether in one p,ob<:, was of special Interest during testing. The individual senWIS within the PIObe yielded accuracies of iO.2"C after calibration. The key oon· c1ulJon from this tesllng was Ihat the bundling of multiple senwrs tOjether had no effect upon the basic fiber optic/GlAs crystal temperature measurement phenomenon. Small degrada· tion of accuracy for the linear array is probably a lesult of I temperature "adient along the calibration well. Key design features of both the thermometlY system and ICCeuorle-s have oomblned 10 make it well suited to the clinical environment. The primary facton for cllnicallpplica· tion such a~ acculacy, multiple point tempenature measule· ment, automatic calibration, and optra tor system con troll monitOring have b<:en more than adequately addressed by the sy~tem's basic design. Othe, lmponant factors rellte 10 how the sensors must b<: handled. brought to the treatment site, and actually inserted. The thermometry ~ystem is relatively 1Illall fOI a multiple prob<: system (19 cm high, 29 em .... Ide, Ind 50.8 em long) so that It can be easily placed in dose proximity 10 the patient. Thc senSOr prob<: length of 1.2 m allo .... s enough distance be· tween systcm and paUent for placemelll of probe, at several diffelent tltatmelll sites. Almost half the length of each pmb<: is covered with a Tef10n sleeve for protcctlon apln~t rough h.andllng. The pmb<: tips. howeve r. are delicate due to thei, small diame ters and muS! b<: handled with care (see Figs. 2 and 3). The clinical requirement for a simple but safe method of sensor insertina into living tissue has been met by the closed end calheter technique. For a single senSOI probe this en"ils the use of a sterUe 19 gauge dOSl'd end ~athe ter and sterile 11 gauge breakaway hYplxkrmic needle (16 puge catheter and 14 puge needle for the linear amy). The catheter h Imt inserted into the necdle to give Itlhe lIillp point and stiffncl-S needed for plOpel placement of the needle/~atheter at the OOr· rCCI Site, angle, and depth. Aft er insertion, the needle Is Iplit and withdrawn leaving the catheter in the tissue. The sensor can then b<: Inserted Into the catheter to implement tempera· tur. measurements. ACKNOWL.;[)GMENT The authors willi to expless their appreciat ion to K. Chun for hill .dvice and technical assistance. 172 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-31, NO.1, JANUARY 1984 REFERENCES [I) D. A. Christensen and C. H. Durney. "Hyperthennia production for cancer therapy: A review of fundamentals and methods," J. Microwave Power, vol. 16, pp. 89-105, 1981. (2] R. R. Bowman, "A probe for measuring temperature in radio frequency-heated material," IEEE Trans. Microwave Theory Tech., vol. MTT-24, pp. 43--45, 1976. [3] L. E. Larsen, R. A. Moore, and J. Acevedo, "A microwave de­coupled brain temperature transducer," IEEE Trans. Microwave Theory Tech., vol. MTT-22, pp. 438--444, 1974. [4] T. C. Cetas. "A birefringent crystal optical thennometer fOf meas­urements in electromagnetically induced heating," in Proc. USNC/ URSJ Symp. (Bur. Radiol. Health), Rockville, MD, 1975, C. C. Johnson and J. L. Shore, Eds. [5] T. C. Rozzell, C. C. Johnson, C. H. Durney, J. L. Lords. and R. G. Olsen, "A nonperturbing temperature sensor for measurements in electromagnetic fields," J. Microwave Power, voL 9, pp. 241- 249, 1974. [6] C. C. Johnson, O. P. Gandhi, and T. C. Rozzell, "A prototype liquid crystal fiberoptic probe for temperature and power meas­urements in RF fields," MicrowaveJ., vol. 18, pp. 55-59,1975. [7] K. A. Wickersheim and R. V. Alves. "Recent advances in optical temperature measurements," Ind. Res. Develop., vol. 21, p. 82, 1979. [8] T. Samulski and P. N. Shrivastava, "Photoluminescent thermome­ter probes: Temperature measurements in microwave fieMs," Sci­ence, vol. 208, pp. 193-194, 1980. [9J D. A. Christensen, "A new nonperturbing temperature probe using semiconductor band edge shift," J. Bioeng., vol. I, pp. 541-545, 1977. {IOJ J. I. Pankove, Optical Processes in Semiconductors. New York: Dover, 1975. {II] D. Gloge, "Bending loss in multimode fibers with graded and ungraded core index," Appl. 0pI., vol. II, pp. 2506--2513, 1972. [12J K. H. Yang and J. D. Kingsley, "Calculation of coupling losses between light emitting diodes and low-loss optical fibers," Appl. Opt., vol. 14, pp. 288-293, 1975. [13] V. A. Vaguine, R. H. Giebeler, Jr., A. H. McEuen, and G. M. Hahn, •. A microwave direct-contact applicator system for hyper­thermia therapy research," in Proc. 3rd Int. Symp. Cancer Therapy by Hyperthermia, Drugs, Rad., NCI Monograph 6,1982, pp. 461- 464. ~ictor A. V~guine (M'82) photograph and biography not available at the time of publication. Dougla.s A. Christensen (M'69) photograph and biography not available at the time of publication. Joe H. Lindley, photograph and biography not available at the time of publication. Thomas E. Walston, photograph and biography not available at the time of publication.