An Adaptive-FocusingAlgorithmfor a Microwave Planar Phased­Array Hyperthermia SystemAlan ].Fenn, Chris J. Diederich, and Paul R. Stauffer

• We have experimentally investigated the use of adaptive-focusing techniquesin the hyperthermia treatment of cancer. Gradient-search adaptive-focusingsoftware developed at Lincoln Laboratory has been implemented on acommercial microwave hyperthermia planar phased-array antenna system at theUniversity of California at San Francisco (UCSF). The system, manufactured byLabthermics Technologies, Inc., consists of 16 independent amplitude/phase­controlled waveguide antenna elements operating at 915 MHz.

In the experiments, conducted at UCSF, a method of steepest-ascentgradient-search feedback algorithm was used to control the hyperthermia arrayphase shifters and focus the transmitted radiation beam at a hypothetical tumorsite in a muscle-equivalent sugar/saline liquid phantom. A feedback probe,embedded in the phantom, measured the resulting electric field (E-field)generated by the antenna array. The measured data indicate a significantincrease in the focal-region field strength with a rapid convergence of theadaptive-focusing algorithm in 10 to 15 iterations. From the measurements, themaximum useful heating depth in thesugarlsaline phantom is estimated for the915-MHz system at about 3 em.

ADAPTIVE ARRAY ANTENNAS are well known fortheir ability to improve, in real time, the per­formance of communications and radar sys­

tems [1-4]. Recendy, adaptive array techniques havebeen applied in the medical field for the hyperther­mia treatment of deep-seated tumors [5-12]. (For anintroduction to hyperthermia treatment, see the box,"Treating Cancer with Hyperthermia," on the follow­ing page.) With an adaptive radio frequency (RF) ormicrowave hyperthermia array, it is possible to con­trol the electric field (E-field) automatically at mul­tiple positions within the target body [5-12]. The E­field radiated by a hyperthermia phased array can beminimized (nulled) and maximized (focused) at de­sired target positions by adaptively adjusting the trans­mit amplifiers and phase shifters of the hyperthermiaapparatus. Multiple adaptive E-field nulls and adap-

tive focusing were experimentally demonstrated atthe State University of New York (SUNY) HealthScience Center on a modified commercial RF hyper­thermia system. In the experiments, gradient-searchsoftware, developed at Lincoln Laboratory, was usedto control the generated E-field of the Sigma-60(manufactured by BSD Medical Corp. of Salt LakeCity, Utah), an annular phased-array antenna systemoperating at approximately 100 MHz [9-14]. Detailsof the SUNY experiments have been reported in anearlier issue of this journal [11].

Subsequent to the SUNY experiments, the samegradient-search algorithm was implemented on a dif­ferent commercial hyperthermia system-theMicrotherm-lOOO (manufactured by LabthermicsTechnologies, Inc., of Champaign, Illinois), a planarphased-array microwave antenna system operating at

VOLUME 6, NUMBER Z, 1993 THE LINCOLN lABORATORY JOURNH 269

• FENN, DIEDERICH, AND STAUffERAll Adnpriw+FOC'using Algoritlmlfor II Mi('fOIllIll~PIII/Illr PI/JIJrd·l1rrnJ IIJpfrlbrrmilt Spfl'l/l

TREATING CANCER WITH HYPERTHERMIA

THE TREATMENT OF malignantcumors is often a difficult task.The objective is to reduce orcompletely remove the tumormass through the use of one ormore modalities, commonly sur·gery, chemotherapy, and radiationtherapy [11. One particular meth·od used in conjunction with an­other modality is hyperthermia[1-6], in which a tumor is heatedto diminish it.

Hyperthermia treatment re­quires a controlled thermal dosedistribution. Typical localizedhyperthermia temperatures thatare necessary for the therapeutictreatment of cancer are in therange 42.5° to 45°C. (The nor·mal temperature of human tissueis 3rc.) During treatment, sur·rounding healthy tissue shouldbe kept :1.( temperamres below42.S°C. The most difficult aspectof inducing hyperthermia witheither electromagnetic or acous·tic (ultrasound) waves is produc­ing sufficiem heating at the siteof a tumor, especially a deep­seated one, without damaging anyhealthy tissue.

Fordecuomagnetic hyperther­mia tre:Hmem, muhiple-antenn:Iradio frequency (RF) or micro­wave hyperthermia :Irrays :Irecommonly used to provide a fo­cused main beam at the tumorsite. A focal region should be con­centrated at the tumor with min­imal energy delivered to the sur-

rounding normal tissue. Becausethe hyperthermia antenna beam·width is proportional to the wave­length, a small focal region sug­gests that the wavelength be assmall as possible. Due to propa­gation losses in tissue, however,the penetration depth decreaseswith increasing transmit frequen­cy. Typically, for noninvasivephased arrays an operating fre­quency close to 100 MHz is rec·ommended for the heating ofdeep-seated tumors and a frequen.cy close to 900 MHz for the heat·ing ofshallow tumors.

A typical clinical RF or micro·\V:lve hyperthermia treatment con·sisrs of several [Wo~hour sessionsspread over a period of weeks.During the first hour of a session,Jnsrrumentation to monitor tern·perature and other vital signs isattached to the patient. Next. forabout 15 minutes, the hyper­thermia equipmem is turned onand adjusted to achieve the de­sired temperature in the tumor.The adjustments to the hyper­thermia equipment can consist ofchanges in the transmit power lev­el (for single-antenna devices). orchanges in both the transmil pow­er level and phase (for multiple­antenna devices). The tumor isthen heated for approximately45 minutes by radiated e1ectro­magnetlc energy to a temper:ltureof 42.5° to 45°C. Note thar, inorder to limit the overall length

of the treatment sessions to tWOhours, only a maximum of 15minutes is available for adjustingthe hyperthermia equipment to

achieve the ideal electric field (E·field) distribution for a particularpatient.

Current clinical operation ofcommercial hyperthermia ph~d­array devices allows limited man­ual control of the array transmit­antenna amplitude and phase.Although this manual trial-and·error method can achieve someimprovement in the E-field dis­tribution, automatic adjustmenttechniques, such as those offeredby adaptive arrays, are desirablebecause of their promise of fast­er operation and bene.r E-fidddistributions.

Several journals have publishedspecial issues 011 the acoustic andelectromagnetic hyperthermiatreatment of cancer [6-81. andvarious articles have investigatedthe widely differing methods forimproving such treatment. For ex­:Imple, many studies [7. 9-12Jhave shown thal phased arrays canbe used to produce improved ther­apeutic field distributions: re­searchers have demonstrated thatphase comrol can synthesize im­proved RF t:ldiation patternswithom adaptive control of thetransmit-array weights, and thattransmit-array weights (definedhere as amplitude and phase statesassoci:Iled with a transmir chan-

"

• FENN, DIEDERICH, AND STAUFFERAll Ad"ptiIJt-FIK/lSlIIg Afgt'rII/'mfor (/ "1i(Tf)l<'fll~ PllI/lilr Pbmrd-Amry Hyp(r(/'tTmill SJ1um

nell can be adaptively controlled[Q maximize the tumor tempera­ture (or microwave power de:li\l~

ered LO a tumor). while mini­mizing the surrounding [issuetemperarure (or microwave pow­er delivered to the surroundingtissue). It has also ~en shownthat ulrrasound receive feedbacksensors [l3] and a pseudoinverse

EkfrrtnU1I. C.A. rein and L w: Bmy. Prindpkr

andPraCliuDfRadiarion Onl'DIooO.B.Lippincolt. Phibddphia. 19Sn.

2. A.W. Guy. ~Hi$IOry of Biological Ef­fe<;tS:mti Medical Applicl.Iiol1$ of Mi­Cl'()\<....,"C Encrgy,~ IEEE Trims. Mkrow.TINo" T«h. MIT·32, 1182 (Sept.1984).

3. J.O...crga=l."ThcEff(l;tof!...ocl.lHy­JlCTlhami.ll Alone and in Combirulionwith Radiation on Solid Tumors. M inCUlm ThmJpy bJ HJpmhcmia aruiIUu1i4licn, Proc 2nd If/I. Symp., 2-4Junt 1997, ed. C. Strdkr (Urban &Sch"''3J'Unberg. Balrimore, 1978), pp.49-61.

4. }. Ovcrg:wd. ~ainical Hypmhermia­An Update, ~ Proc. 8th 1m. Congrm ofRmJilzrion nr-rrh 2. juL /9$1. eds.E.M. Fielden.j.E Fowler, }.H. Hendry,and D.O. &ou, p. 942.

5. S.B. Field and j.W H:md, eds., All In­tl"rNilmion 10 tiN Pnuticillkpms ofClin­icd H,pmJxrmilz (Taylor & Fr=cis.,london. 1990).

6. ~SpeciaI Issue on Can«r Therapy byElCO"romagnctic Hyperthermia,~ j. Mi­cn>u.l ~160une1981).

7. ·Speciallssuc on Phased Arnys for Hy­perthermiaTrcl.[mcntofCancer,· IEEETmns.. Microw. TINory T«h. MlT-34(May 1986).

8. ·SpeciaI Issue on Hyperthermia andCancer Thcr.apy; IEEE Tmm. Biomcti.till. BME-31 O:m. 1984).

9. D. Sulliv:m, ~Thltt-Dimcnsional Com­pUler Simulation in ncq, Ikgional Hy­pCTthermia Using the Finite-DifferenceTime-Domain Me1:hod.~ IEEE Tmm.Mirrow. Tlxory T«b. MTf-38, 204(Feb. 1990).

10. R.B. Roemer. K. Hynynen, e.johnson,and It Krc:ss, ~Fttdback Control andOplirniz:nion or Hyperthermia Heat-

pattern synthesis method can beused to geneme multiple focalpoints wirh a phased-array hyper­thermia applicator [14, 151. Oth­er studies [16, 17J and several con­ference articles [18-20] haveinvestigated the theoretical bene~

fit of using near-field adaptivenulling [21-24] with noninvasive

auxiliary dipole sensors to reduce

ing P;tterns: Present StatuS and FutureKecds.~ I££E Eighzh Alll/udCol/! Eng.Meti. BioL 5«.. p. 14% (Nov. 1986).

11. P: Wust.j. Nadobny. R Fdix, P: Deufl­hard, A. louis, and W.john, ~Strate­

gies for Optimized Application of An­nular-Phased-Amy SystCn1S in Oinic:aJHyperthermia," InL}. H,pmhermilz7.157 Oan.-Feb. 1991}.

12.j.W. Srrohbehn, E.H. Curtis. K.D.P;u15en, :md D.R Lyndl, ~Optimw.­

don of the Absorbed Power Distribu­tion for an Annubr Phased-Amy Hy­pcnhermia System: Inr. f RmJiilrUmOnroL BioL Phys. 16, 589 (1989).

13. E.S. Ebbini. H. Wang, M. O·Donndl.and CA. Cain, "Acoustic: Fcedb3dt forHyperthermia Phased-Array AppliCl.­tors: Aberration Correction. MotionCompe~tion, and Multiple Focusingin the Pre:scnce ofTISSUe Inhomogene­ities,~ Proc. IEEE /99/ UlrrllSl)n. Symp.2,1343 (1991).

14. E.S. Ebbini and C.A. Cotin. -Experi­mental Evalu.uion of a Prototype Cy­lindric:aJ $a:rion Ulu250und Hypmha­mia Phased-ArI'1llY AppliCl.lor," IEEETroffS,. Ulmwn. !YrrMkm: Freq. Con­l7'()/38, 510 (1991).

15. E.S. Ebbini:md C.A. Cain, -Mulripl.....Focus Ultrasound Phased-Array P;ucmSynthesis: Optional Driving-Sigrul Dis­tributions for Hyperthermia; lEE£Tmm. Ulmwn. !YmNJ«rT. Frrq. Con­trol36, 540 (1989).

16. A.J. Fenn, ~Applic.ation of Adap­tive Nulling to Electromagnelic Hy.perthermia ror Improved ThermalDose Distribution in G-mcer Thera­py: Trdlllimi Heport 917. LincolnLaboratory (3 July 1991), OTICItAD-A241026.

17. A.]. Fenn and G.A. King. "An Adap­tive Radio-Frequency Hyperthermia

the E-field intensity at selectedpositions in a target body whilemaintaining a desired focus at a

tumor site. In particular, multi­ple adaptive E~field nulls wereused to show a theoretical reduc­tion in hot spots for a homoge~

neous elliptical phantom targetsurrounded by a water bolus andhyperthermia ring array [18-20].

Pha.s«\-Array System for ImprovedCancer Therapy: Phantom TargetMeasurements," T«hnkill Report 999.Lincoln Ltbor:ttory (19 Nov. 1993),ESC-TR-9J..299.

18. A..j. Fenn...Foo....scd Ncat-Fidd Null­ing for Adaptive E1=magnetic Hy­perthermia Applications, ~ /99/ frog.£larromtZgn. &s. SJmp. Prtx. Cam­bridge. MA, /-5j",/y 199/, p. 393-

19. A.j. Fenn, ·Adaptlvc: Hyperthermia forImproved Thermal Dose Distribution."in RmJilztion ~ilrrh; A Twemiezh­Crnt:lry Pmp«rive I: CAngrm Absm:(TS,7-/1 juJ, /991, cds.j.D. ChaplTU1\,We. IXY.-cy, :md G.F. Whitmore (A<:­adanie Press, Sm Diego. 1991), p.290.

20. A.J. Fenn, -Noninvasivc:Aaaprivc: Null­ing for Improved Hypmhermia Ther­mal Dose Distribudon,- IEEE Eng.M«l. BioL Soc. InL Con! 13, 976 (31Oct.-3 Nov. 1991).

21. A.J. Fenn. -Theory and Analysis ofNeat-Fidd Adaptive: Nulling, ~ /986kiIomilT Con! on Sigmsls, Sptmu Qlld

CompulnS (ComputCT Society Press ofthe IEEE, Washington, DC, 1986), pp.105-109.

22. A..J. Fenn, "Ev:aluadon of AdaplivePhased-Amy Antcnrut Far-Field Null­ing Performan« in the Near-Fidd Re­gion," lEEETmns.. Mtnr1JilS P'roJH1g. 38,173 (1990).

23. A.J. Fenn. H.M. Aum:mn, EG. Will­werth, and J.R. Johnson, ·Focused.Near-Fidd Adaplive Nulling: oped­mentailnvestig;ltion." /990 IEEE AIl­fellllllS ProJH1g. 5«. IIIL Symp. Dig. I,7-11 May 1990. p. 186.

24. A.J. Fenn. "Anal)'liis or Phasc:-FocuscdNcar-FiddTesling for Mulliphasc·Ccn­ler AdJpdvc: Radar Syslems: IEEETm1lJ. Anullluu Propng. 40, 878 (Aug.1992).

10l~MI 6 .~MB!~ 1 1993 r~( 1IlCOli lIBO~ll0~1 JOURlll 271

..• FENN, DIEDERICH. AND STAUFFER

A" AtiDptlly.FtKllfi,,: Alpnrhm ftr" /tflC'l'O_~YPbz.."r PlNtHtI.A....., H,pnThnm'6 jl'lfflI

915 MHz. Expcrimc=nts were conducted at the Uni·versity of California at San Francisco (UCSF) Radia­tion Oncology Depanmem in which the algorithm

was used for comrolling !.he hyperthermia array phaseshifters to focus the transmim:d radiation beam :11 a

hypothetical tumor site in :l sugarfsalinc liquid phan­tom (dielectric losses of3 dB/em). The UCSF experi­ments are the subject of this article.

~eT21 major differences exist bc=rween the SUNYand UCSF experimenrs. In panicuJar, the hyperther­mia equipment uso:l in the two experimenrs had dif­ferem transmit frequencies (tOO MHz versus 915MHz) 2nd array geometries (annular versus planar).

BecauS(: of the difference in transmit frequencies. theSUNY experimcnts investig:ncd thc treatment ofdcep­seated {Umors (depth gre:llcr [han about 10 em) whilethe UCSF experimentS wcre concerned with shallowtumors (depth less than about 3 em). In addition. theSUNY experimenrs investigated the use of adaptive

nulling as well as adaptive focusing while the UCSFexperiments considered just adaptive focusing.

Background

Microwave antenna applicators have frequently beenconsidered for inducing localized hyperrhermia 10

superficialwmors located 011 lile chest wall and headand neck regions. For single~aperrure applicators, how­ever, the power-deposition patterns are often limitedto effective heating well within the apenure bound­aries [lS-18). (Note: The power-deposilion patternsare also referred to as [he specific absorption rate(SARl, which is equal to Yza 1£1~/p, where £is the E­field, G is the elecuic.,1 conductivity of the [issue, andp is the tissue density.) With 915-MHz microwaveradiation, for example, ::t waveguide applicllor has aheating deplil limited to less Ihan 3 em and l::tteralheating dimensions of 3 [0 5 em-insufficient totreat many tumors [191. The he<tting depth is limilcd

16-channel microwave transmitter Sugar/saline liqUid phantom

FIGURE t. The Microtherm-IOOO (manufactured by Labthermics TechnologIes, Inc., of Champaign. III.), a 915­MHz microwave planar phased-array hyperthermia system used in experiments at the University of Californiaat San Francisco (UCSF) Radiation Oncology Department.

272 UI Illl(:Oll 1110111011 ""'Ul 'OUIilI. "'11I11' 1M3

• FENN. DIFI)I:HICII. ANI) STAun'ERAll Adupwlr-FtKlIsinz AIJOrtf/"n fPr II MltTVU/", ... Pumar Phasn/·ArTIrJ HJpmhrnmll SJfUM

15.2cm

FIGURE 2. The Microtherm·l000 system: (8) footprint and(b) cross section 01 applicator.

••

...

Coaxialfeed lines

c

Distendablebolus

(deionized water)

K G

(.J

Waveguides(16)

o

(b)

Additionalhigh.dielectric­

constantmaterial

/

,

Applicatorhousing

I~r+-r---,I-------,

30m

I

Aperturesurface

\ ....

Outerapplicatorhousing

Eu

'"

field probe used in the UCSF measurcments was :tshort dipole with a semiconduaor diode detector(51-54). (Note: Thc Microtherm-IOOO currently doesnot supply [·ficld probes to monitor clinical hy~r­lhcrmia treatments; however, as the currcnt measure­mems indicatc. an E-field probe could be added toIhe system 10 provide feedback signals to the adaptivealgorithm. In theory. with 16 independent transmitchannels. the [-field rndiation pattern can be con­trolled at up to 16 poims [in the radiation field! byllsing a feedback signal measured at each point.) In

to only about 3 em because the dielectric loss of915­MHz radiation in lissue is typically 3 dB/em. Fromprevious experiments, researchers have found that apredictor of good lOCI! healing comrol is to use the50% iso~SAR coverage (laterally and with depth) ofthe tumor (20].

Several mtthods for improving the SAR heatingpatterns of single-aperture hyperthermia devicts areunder investigation. For example, microwave-absorb­ing saline-filled boluses can shape the hearing panernof a waveguide applicator, as has been demonstratedin Sherar et al. [211. Mechanically scanned micro­wave applicators have also been used to shape ,hepower-deposition pattern (19, 22, 231.

Another method of improving the SAR he'Hingpattern is to use muhiple.aperture arrays (24-431.Researchers have found that phased-array applicatorsan be used to shape the power-deposition pallernproduced by planar (19. 22. 44. 45J and conformalarrays [46,471. And stud its 10 improve penelrationdeplh with phased :trrays by COlli rolling the phase andamplitude ofeach array clement have been conducted[44,45,481,

In our experiments, we: have used the Microthenn.1000-a hyperthermia microwave system with 16alllennas (Figure I) [491. By transmitting close to thepatient with a planar array, the system is able toobtain superficial heating over large (approximately15 x 15 em) as well as small (approximately 3.5 X3.5cm) areas. depending on the array amplitude illumi­nation function. Figure 2 shows the 15 x 15-cl11planar array wilh its 16 square waveguide eIemenlsoperating al 915 MHz. The 16 independent variablepower amplifiers drive: Ihe waveguides, and each ofthe 16 active channels has an electronically controlledvariable-phase shifter 10 focus the array. A cool~water

bolus between the: patient and the phased arrayprevents excess healing of the skin surF.1ce. The bolusis filled with circulating deionized water, which hasa very low microwave propagation loss of abom0.3 dB/cm.

Previously, UCSF had evaluated (in experimentswith phantom materials such as deionized water andmuscle-equivalent liquid, and with animals) a system·atic-se:lrch, iterative focusing algorithm for phased.array control of the Microtherrn-1000 [501. The E·

,OlUWI8 XUlurl2 IIU '~llINCOlW l1101l1011 ~OUUll 273

• "-ENN, DI£DERICII, ANI> STAUFFERAll Adaptit~FtKUJ;"X Alxoritbmfor" MltTtIWQw Plmlllr f'I'IIsnl.ArrIlJ NJprrtJ"rmIJl SJSlmI

Hyperthermia array

~

Receiver

Focus-",,,,r

Feedbacksignal

"'-Y'....T urnor

E·field sensor

~

R'source

Weightcommands

(amplitude/phase)

ISignal processor

Adaptive·focusingalgorithm

FIGURE J. Adaptive-focusing hyperthermia system concept.

this aniclc. we investig:nc an alternative Clndidatealgorithm for the ad:tptive focusing of the Microthcrm­1000. Developed by Lincoln l...:tboratory. the com­puter algorithm uses a gradient search based on themaximization of the signal that is received by an E­field sensor positioned within a tumor.

Theory

Adnptiw.Focusing Hyptrtbamin SyJum COtlctpt

The concept of an adaptive.focusing hyperthermiaSYSlem is shown in r-igure 3. To generate the desiredE-ficld disrribulion with a clinical ad:tptive hyper­thermia system, a field probe is positioned as closelyas possible to the lllmor site and the hypenhermi:tarray is focllsed to produce the required field intensityat that site. The probe provides feedback from whichthe: adaptive.:trray weights W

IICOln be adjusted to con­

trol the: amplirudes A and phases ~ of the individualantenna elemenfs such that the energy received 3( thetumor sile is maximized. (NOIe: Although Figure 3shows only one field probe. an array ofprobes may beu=l.)

Adnptiz~Tmmmil·A"ay FontmlllliolJ

Consider a hy~rth~rmia array with N am~nna e1e­m~nlS (N '" 16 for the Microth~rm-l000 system).

The input signal to each of the N array elements isobtained from the amplitude and phase:.adjusted sig­nal distributed by a power divider network. The num­ber ofadapt ive chan ncls is assumed to be equal 10 thenumber of transmit antennas N.

Commonly, the weights are constrained to delivera required amoul1l of power to the hyperthermiaarray or the tumor, For simplicity in the experimentaladapdve-hypcflhcrmia-arr.lY comrol software, we con~

strain the weights such that

",'

where lw"l is the transmit-weight magnitude for the11th adaptive channel and Kis a constant. To generatean adaptive phase focus. a gradient-search algorithmcan be used to control the transmit weights (phaseshifters).

Cmditm·Srl1rch Algorithm

Gradient-search algorithms are commonly used inadaptive-array appliCltions, particularly when thechannel-Io-channel correlation of the antenna de·mems Clnnot be Cllculated or measured. W'ith a gra­dient search, only the received power at the E-ficldprobe(s) is measured and used as a feedback signal to

274 Illl-etlllll.OllICllf .IO.....l l'Ol••t, ".Itt 2 ltt3

• FENN. D1F.DERICI-I.AND STAUFFERAn Ad4plll't--FtH'UJI"1 A/pmll"njiI~ il .lll('1'()U!4L'( (>141Ulr PhmnJ-ArrllJ HJI'"l'OffIm# Spum

the algorithm. Ekc3u~ the Microthenn-l 000 systemat UCSF measures only me received E-field power. itis appropriate to consider a gradient-search algorithmfor the application.

GradienBearch algorithms control the transmitweights iter.ttivdy to maximize (focus) the microwavesignal received by the field pro~(s). Transmit-arrayphase: shifters are adaptively changed in small incre­ments (the process is called dithering) and the re­ceived power at the pro~(s) is monitOred to deter­mine the phase settings that will increase the powermost rapidly to a maximum. A wide variety of gradi.ent searches exiSts [55-60); in our work we have useda standard method ofsteepest ascent. The mathemati-

cal formulation for the method is straightforward155.61]; a detailed description of the formubtion inthe context of hyperthermia is contained in Refer­ences 9 and 62.

Spurn Cons;d"al;ons

Figure 4 is a block diagram of an adaptive hyperther­mia system controlled by a gradient-search algorithm.Transmit weights W1j' ...• w"}' ... , wNj at the jthiteration are shown at the [Op of the figure. Thetra.nsmit phased-array al1tcnna induces a voltage acrossthe terminal of the ;,h receive field-probe antenna.(Note: Figure 4 assumes that the system lIses an arrayof Naux field probes.) For any given configura.tion of

Control

nTransmit

• •• N

Transmitweights(ith state)

Transmitarray

i=1.2•...• N.U •

ith receive probe (total of N'UJ probes)

Gain -..Adjust a

Receive,compute,and control

Received power p

Feedback signal (PI)~---'--~

Computeweight search

directions

Computeweight updates

Control

Adaptive transmit weight (j + 1)th stale

FIGURE4. Block diagram of gradient-search algorithm for adaptive-focusing hyperther­mia system.

,Ollfll( 6 .UU1111 1113 r~llI\COlllllalirOAl JOUlUl 275

• FENN, DIEDERICH, AND STAUFFERAll Adllf'ltl'f'-Forusmg Algorill"" for II />1,rrowll''l! PI",,,,, Pbmld-Arr"J HJprrtlH'rmlll SYS/lm

..

FIGURE 6. Amplifier circuit for E-field probe.

,-----'\/W------,-,----\ +

High-resistancesensor leads

depending on rhefewer chan abour 15 iterations,gradieOl~search Slep size {j,1/J.

Materials and Methods

The hyperthermia phased~array system used in thesemeasurements is rhe Labrhermics MicrOlherm-1000planar phased-array applicator (Figure I). The systemconsists ora 15 x 15-cm planar array with a 4 x 4 gridof 16 uniformly spaced square waveguide elementsoperaring ar 915 MHz (Figure 2). Figure 5 shows aphorograph of Ihe planar phased-array applicator.Because of the water bolus covering the applicator,the acwal waveguide radialOrs or {he array are nOI

visible in dle photograph. The aperture dimensionsof the individual waveguides are 3.8 X 3.8 cm, andthe waveguides arc filled wirh a high-dielectric­constant material to provide maximum radiationar 915 MHz. The waveguides are linearly polarizedwith the dominant E-field component aligned with

they-direction (refer to Figure 2).The 16 individual waveguides are driven by 16

high-power ampliFiers with up to 35-W average powerper channel. The power applied to each element iscOlllrollable in 10% increments of the adjustable mas·ter power leveL E.1.ch of Ihe 16 active channels has anelectronically controlled variable phase shifter (8 bits,0° w 360°) for focllsing, the array. The phased~array

applicawr can be driven in incoherent mode (with 16independently operating sources with frequencies ncar915 M Hz) or coherent mode (with a single-frequencygener.uor :md independent control of the relative phaseof each aperture to within 1.5°). Invasive Luxtronfiber~optic temperature sensors [63] are included wirhthe Microtherm-l000 system to monitor the lem­

pcratures obtained during tre:ltmenc.To prevent excess heating of a patient's skin sur­

F.tce, the Microlherm-IOOO uses a cool-waler bolusplaced between the patient and the phased array. Thebolus consisls of deionized water, which has a very

low microwave propagalion loss, contained within ;l

flcxible membrane Ihal can be expanded outward to 6to 8 cm in from of the applicalOr fuce to improve themicrowave coupling berween Ihe applicator and tis­SllC. Thc bolus rcmpcrarure is conrrollcd by circulat­ing cooling water through a thermal-conduction plas~

ric-tubing heat exchanger located in the outer cdges

soooDiode to measure 500 nmagnitude of E-fjeJd

60 knfm

/

the transmit weights, each weight is dithered by asmall amoufll in amplirucle and phase, and rhe re­ceived power :H the irh probe is stored in the com­puter for calculation of rhe lora I received power fromthe array of Naux probes, the amplitude and phase

search directions [62], and rhe updated (j + Ihhlrallsmil-weighr configuration. Olle rr.msmit weightis dithered with the remaining transmit weights intheir jlh state. The process continues unci! Ihe(j + l)th weight configuration has converged. Basedon other measurements (10, 121 nOi shown here,convergence for adaptive focusing occurs typically in

FIGURE 5. The 915-MHz microwave planar phased-arrayapplicator used in the Microlherm-I(XXI.

276 1~IIl~ClllN lUllU10RI.lOUR~1l YOlUtH 6 ~UIlUR l 1993

• FENN. DIEDERIC.... AND STAUFFERA" Ad'ipliw-Focusiltg Algorithm jor 'I 1l1ilTOlII,wr PllIIUlr T'!JlIJN(-ArrilY HyprTlh~rmi'l Syslrm

16·channelphased-array antenna ---

Sugar/saline liquid ---­phantom

Dipole probe

:..,;:.~-- Probe (if, Y, z)positioner

FIGURE 7. Dipole probe, probe (x, Y, z) posilioner system, and the muscle-equivalent phantom lank.

of the bolus housing. To improve the cooling capacityand transient control of the bolus in the Microtherm­1000, researchers at UCSF replaced the heal exchangerinside the bolus with a closed~circllit pumping sYSlemthat circubres remperaulre-controlled deionized wa­ler directly through the bolus compartment in serieswith a heat exchanger mounted in a temperature­regulated water bath.

The E·field probe [51, 52. 54J used in our mea~

surements was fabricated at UCSF and integratedwilh Ihe Microtherm-I 000 system. The E-field sen~

sor is a Schottky detector diode (Hewlett-PackardHP-3486 with an olller diameter of 1.9 mm) with thediode leads arranged to form a dipole a/Henna of I emlengrh. The sensor is coupled to an amplifier circuit(Figure 6) via high-resislance lead material to mini­miz.e perturbation of the fields. (Notc: Manuf.1cruredby Holaday Industries. the lead malerial has an outerdiameter of 1.2 mm, including the outer insulation.)The rectified DC OutpUt voltage is a fUllction of theE-ficld squared in the orienrarioll of the dipole leads.Figure 7 shows the dipole probe hardware and (x. y. z)positioner. A 12~bit analog-to-digital (AID) convertersamples Ihe E~ficld probe signal.

In our experi melltS, we used a liquid Illuscle-equiva-

lem phantom (sugar and salt dissolved in deioniz.edwater) to approximate human tissue. The electricalproperties of the phantom (a ~ 1.38 Siemens/m andrdative permittivity £t = 54.7 (641) were similar to

that of high-water-colllent Ilssue at 915 MHz (a =

1.28 Siemens/m :md £, = 51.0 [65J). The liquid phan~

tom was contained within a 50 X 50 X 22-cm-deepPlexiglas tank, which is shown in Figure 7.

The UCSF hyperthermia array is controlled andmonitored by an MS-DOS-hased personal compurersysrem with software that implements the gradienr­search adaplive-focusing algorithm. The software, de­veloped by Lincoln Laboratory and integr::ued byUCSF with the l..abrhcrmics SySlem, allows the opera­lOr to run an adaptive-focusing algorithm chat usesthe OUtput power of the E·fidd probe as a feedbacksignal during rhe gradienr se:lrch. The adaptive~focus­

ing algorithm uses only phase control 10 maximizethe E-rield at the probe's position.

Measured Results

R~vi~w of£arliu Investigations at UCSF

Because of inter-element mlltual~collpling effeclS inthe phased array and because of the variation that can

~OlUW( 5 Ulll!l ,. liU ,~[ lINCQL~ 1I8Q~UQRI JQUUIl 2n

• FENN, DIEDEkJCH, AND sr,\UI'FmtAll Affllplil't-Focming Algoridmifir " M,rrtJU'II/V PI,III"r I'IJ1lsr,I-AmIJ HJprrtlJrrmltl 5pum

17 ,----.-----..,."'qr--,--;rr--,----,

13

9

E~~

5~•~

"

(.)

7 ,-------,----....---,------,

5

3

-5

0.1

ox-distance (em)

5

(b)

10

FIGURES. Measured E-field radiation pattern fort he Microtherm-1000 planarphased array focused at (a) z = 7em in thedeionized-water phantom (propa­gation loss =0.3 dB/em), resulting in a well-focused beam; and at (bl z :::4 em in the muscle-equivalent liquid phantom (propagation loss =3 dB/em),resulting in a beam that is not centered at the desired location. In theexperiment, the UCSF systematic iterative-search algorithm was used tofocus the array. (Note: The radiation pattern contours are shown in 10%power intervals and the location of the focusing probe is indicated with a red"+"sign.)

be expected in phamom m:uerials and humans, it wasnot possible to select II priori the correct phases to

foclis rhe Microtherrn-IOOO 149J. [n addition, re­search was hampered by an inability to calibrate orpreset the MicrOlherm- JOOO's individual phase shiftersaccurately 10 absolute values. Anempts at using theo­retically selected ph:tses th:\{ were applied to eachclement produced, at best, marginal results in a water

ph:lI11om.To compensate for these problems and to accom­

modate expected tissue heterogeneities in vivo, a phaseselection, or optimiz:nion. scheme was developed at

278 THlllNCOlN UIORIIORI JOU~NIL 'OlUlI15 _UWIII' 1993

UCSF to select the phasing of each transmit channelireralively for the maximization of the measured SARat the desired location. A computer/soFtware systemlinked to the Microtherm~IOOO turned on individual

c1emclHs of the phased arr.ty in a predetermined se·quence :lnd adjusted the phasing iter:ni'lcly until amaximulll SAR was obtained.

In the process, clemelH F (see Figure 2) is turnedon ;It :t 0° phase with :til other clements turned off.

Next. element K is turned on and adjusted iter.tri'ldyin phase (from 0° to 360° in J00 increments) toobt:tillthe maximum SAR reading. ElemenrG is then

• FENN. DIEOEIUCII. AND SfAUFI'ERAll AdJ'pn.,.FonnJJlg Algoml"" fi"" Mirroll'l1l~Plil/lltr PI",~".ArraJ H..rrrrtINTn"" SJ5um

Pig thighmuscle

Multisensortemperature probe (16)

Applicator

Trial focus points

• 2.Scm deep

• 4.Scm deep

in Figure 10(3) indic:ues a maxImum penetrationdepth of aboUl 1 cm (for a 50% SAR) on the centralaxis. This result is consistent with a microwave propa~

gation loss of3 dB/cOl. Figures IO(b) and IO(c) showthe measured SAR palterns for a focus at z'" 2.5 and4.5 em, respectively. For the focus at z '" 2.5 cm, themeasured ma.-cimum penetration depth (on the prin­cipal axis) is 2.3 em, and the beun peak is at z'" 1 cm.For the focus a[ z = 4.5 em, the penetration depth isonly 1.9 em. and the beam~ is at the surface ofthe target.

Intuitively. one would expect the Z= 4.5-cm focusto penemue more deeply than the z '" 2.5-cm focus,but instead the measurements indic:ne a reduction inpenetration depth when the focus is increased beyondz'" 2.5 cm. This discrepancy was the principal reasonIh:tt UCSF was interested in comparing the Lincoln

FIGURE 9. UCSF e~perimenta' test setup tor in vivo mea·surements 01 a pig thigh (propagalion loss = 3 dB/em).Note thai temperature measurements are taken at numer­ous sites by 16 multisensor temperature probes spaced1 em apart. Each 01 the 16 probes measures the tempera­ture at 7 diHerent vertical locations spaced 1 em apart.Thus temperalures within the pig thigh are measured on al-cm x l-em grid at 112 sItes.

rurned on and similarly adjusted until a new maxi­mum SAR reading is obt:lined. This process comin­ues for the remaining e1emen15 in the following se­quence: J, B, 0, . C, E, L. H, I, A, I~ M, and D.During this systematic iterarive·~arch process. thepreviously tested elements are kept on at their opri.mi7..ed values while the n~t element in the sequenceis turned on and ph~ adjuSted to maximize theSAR. The entire procedure, from the adjwting ofdemen! F through 0, takes approximately 1 min tocomplete. Experimen15 wing this phasing approachhave bttn conducted with different marerials; deion­ized-water, tissue-equivalem liquid phantoms, and ap;g<h;gh [50J.

With the UCSF systematic iterative-search algo­rithm, it was possible to focus the Microtherm-lOOO:n depths of 7 em in the deionized·w~uer phantom, asshown in Figure 8(a). (Note: The measured E-fieldradiation pattern COntours in the figure are given in10% power imervals.) Clearly, maximum radialionoccurs in the deionized water in the viCinity of thedesired focus. Based on the 50% SAR value, [hemaximum penetration depth is 15 em in the deion­ized·water medium. Using the same algorithm, simi­lar focusing attempts in a muscle·cquivalem phantom(Figure 8(bJ) did nOI generale a useful focused beamat deplh but produced, in essence, a collimated beamwith a beam peak effectively .11 the surface of thephantom. The data in Figure 8(b) indicate a maxi­mum penetration depth of 3 cm. based on the 50%contour, which reaches z '" 3 cm.

The microwave propagation loss in deionized wa­ter is 0.3 dB/cm. whereas the loss in the muscle·equivalent phantom is dose to 3 dB/cm. Clearly thehigh propagation loss in muscle tissue contributedgreatly to the system's inability to focus at the desireddepth. Nevertheless, a 3-Clll penetration depth is stillpotentially useful for certain shallow tumors.

The efficacy of the UCSF systematic iterative~search

algorithm was also evaluated with a pig thigh in whichan implamed SAR sensor was used [0 oplimize thephase. Figure 9 shows the in vivo experiment.:ll ~tupwith trial focus posilions (at z '" 2.5 and 4.5 cm)indicted by the shaded circles located on the centralaxis. In the incoherent mode with the 16 waveguidesoperating indepcndend)" [he measured SAR pattern

'l'lll..f' U ....U1 IttJ 1~lllltllUlll'OI"'OIIJOIIlIl 279

, .

• FENN, OIEDER1Cl-I, ANI) STAUfFERAll AdJlptn~-ForrlsillgAlgtJTirhmfo,,, /l1irroll'fl/" PltJmlT I'IJllud-Aml) 1l,J>rrtlJrnw{/ S,iU/Il

FIGURE 10. Measured specific absorption rate (SAR) forthe Microtherm-l000 planar phased array illuminating thepig thigh of Figure9: (a) incoherent mode with all elementsincoherently driven, (b) focused at z = 2.5 cm, and (c)focused at z = 4.5 cm. For parts band c, the UCSF system­atic iterative-search algorithm was used to focus the array.(Note: SAR Is defined here as c!:lT/t!J, where c is the spe~

cific heat, and t!.T is the change in temperature over thetime interval t!J =30 sec. The location of the focusingprobe is indicated with a red "+" sign. )

New MtllSlfrtlflmfs at UCSF

UCSF and Lincoln Laboratory began collaboratingin June 1992 to implement an adaptive-focusing,gradienHearch algorithm with the Microtherm-I 000.Experiments were performed wi[h the deionized-wa­ter and the musc1e~eqtli~lentliquid phantoms. Phasefocusing was attempted III the central axis of theMicrotherm applicator at depths of6 and 8 cm from

rhe lowest ridge on the applicator housing with anaddirional 3~cm path length in deionized water from

the apertures to this point.Measured E-freld pacterns (Figure 12) with the

focus set at 8 em indicate that both optimi"l.:ltion

Labor,Hory gradiclH-search algorithm wilh the sys­tcmalic iterative-search algorithm that had beell used.

In Figure 11, the measured SAR is compared withthe temperature distribution (or the 2.5~cm~focus case

(Figure IO\bJ), and the twO sets of data are consistent.That is, the peak temperature value occurs near thepeak of the SAR distribution. (Note: SAR is definedhere as cA TIAI. where c is the specific heal, and ATis[he change in temperature over the time interval 6.1.)

From the results shown in Figure 8(a), we con~

eluded th;H the systemaric irer.nive-search algorithmcould be used to focus the Microtherm-IOOO in a10w~loss deionized-water phantom. With the musclephalllolll and pig thigh, however, the algorithm didnOl result in useful focuses (although it did illustrate apossible improvement in penetration depth). Thusone question that remained concerned the effective­ness of the systematic iterative-search algorithm. In

particular, we were eager to investigate whether benerfocusing in a lossy medium could be obtained with anadaptive oprimi7A1rion algorithm, as had been consid­ered previously [7,10, 12J.

32

2 3

2 3

I I

-

~

I

a

a

I

-1

-1

I

+

-1 ax-distance (em)

(0)

x-distance (em)

I')

x-distance (em)Ib)

-2

-2

-2

-3

-3

-3

I

O·~--""""""O.4

~.2~02-----1--::

-

5

6

5

a

6

280 !HI LI~COl~ lUOUIO~' JOUUIl VOlUMt6 MUMUR 2, 1993

• FENN. DI£DERIC.... AND STAUFFERAll Ad'lpull(-FtKusillJ, Algorithm for II Micro/IN/I., PftlllJlr Phasrd-ArraJ f/Jputlunnill SJStt'1II

o

6(.) (b)

5

E 4q~

!!-£ 3C.•~

" 2 0.6

x-distance (em) x-distance (em)

FIGURE 11. Comparison of (a) measured SAR and (b) temperature distributions (given in degrees centi­grade for steady-state conditions aher 10 min) for the Microtherm-1000 system focused at 2.5 em in the pigthigh of Figure 9. In the experiment, the UCSF systematic iterative-search algorithm was used to focus theMicrotherm-l000. (Note: SAR is defined here ascAT/M, wherec is the specific heat, and AT is the change intemperature overthetime interval At = 30 sec. The location of the focusing probe is indicated with a red "+"sign.)

schemes-the UCSF systematic iterarive~search fo­

cusing algorithm and the Lincoln Laboratoryadap­tive gradient-search focusing algorithm-work simi­larly. Both algorithms result in well-defined focusedbeams :11 the desired depths. The current measure­lllents do not consider field mappings in other verti­cal and horii'-Omal planes. Such additional mappings:lre necessary to characterize the locations and m:lgni­mdes of the E-field siddobes and possible grating

lobes.It is interesting to nOte that the radiation patterns

in Figure 12 exhibit a z.-dependellt ripple with spac­ing approximately equal to 1.8 cm. We suspected thatthis ripple was caused by a standing-wave patterncre:necl by incident and renectcd waves in the phan­tom water tank. By assuming a dieleCtric constant of80 and an electrical conductiviry of 0.19 Siemens/m,we calculated the wavelength of915-MHz microwaveradiation in deionized water as 3.6 cm. Thus theripple spacing was equal to one-half the wavelength,as was expected with a standing-wave pattern.

Measurements in the muscle-equivalent phantom

were performed at a depth of 4 cm for three cases ofamplitude illumination: equal illumination applied at

100% of the selected power level of 5 W (Figure 13),adjllstcd uniform illumination with premeasured SARamplitudes for each element (Figure 14) [49], andinverse~tapered illumination with the application ofdouble the power to the outer elements (Figure 15).These radi:uion-parrcrn measurements arc similar tothose obtained in earlier studies with the UCSF sys­tematic itcrativc~search algorithm [50]. Note that (hereare only minor differences between lhe three radia­

tion patterns of Figures 13 through 15. It is very likelythat the four-element group at [he center of the arrayis the dominant contriblllor to the radiation patternshape.

Lastly, Figure [6 shows the convergence of thegradient-search steepest-ascent algorithm for the caseof the 4-cm focal depth. The measured E-field at thefOClIS has been ploned on the y-axis in power indecibels by computing 1Olog(p), where p is the value(measured power) of the AID convener. In the figure,

the focal-probe power increases by 17 dB over the 15iterations with the data indicating convergcnce inabour 7 iterations.

In the experiments. each phase shifter was con­trolled by an 8-bit D/A convener that had 256 states

'OlUIiE 6. NUWBIR 7. 19U 1"1 tlNCOlN lJlORUOU JOUhll 281

• FENN. DIFDERICH. A,"l'D SfAUFFERIt" AJapuw-ForwIIlf Allomh",fOr u M,rrtlU.'flIV PiJlllnr Phouni-Am'J HJpmJonmu, Spum

16

11

E

'"~0.•~" 6

1

-10 -5 o,-distance (cm)

(.j

5 10

105o,-distance (em)

(b)

-51-10

16

11

E

'"~0.•~,

\J- 6

0

FIGURE 12. Comparison 01 measured two-dimensional radiation pallerns of theMierotherm-1000 with focal depth of 8 cm in deionized-water phantom: (a) UCSF sys­

lematic iterative-search focusing algorithm and (b) lincoln laboratory adaptive gradi­ent-search focusing algorithm. (Note: The radiation patterns are shown in 10% powerlevels.)

282 I~I -eDt_LUNI"n Hllhll 'f'It11l(' '11111., 1993

• FENN, DIEDERICtI, AND STAUFFERA" NWflllY-F(I('IlJUlK A'Koml.",JPr" M,nrlu'/l/Jt' r"lIInr P"il~d.Arr".r H..y/'",IJtTm,Il Sp","

100 100 100 100

y

100 100 100 1001-

100 100 100 100

100 100 100 100

(.)

7

~5 0.1~

<i• 3~

.:.1-10 -, , 10

x-distance (em)(b)

FIGURE 13. Effect of array illuminatIon on radiation panern for the case of Uniform illumination: (a) array illumination (inpercent of the selected power level of 5 W) and (b) resulting two·dimenSlonal radiation pattern (in 10% power levels) of theMic,otherm·tOOO using Uncoln Laboratoryadaptlve-focuslOg algorithm with 4-em focal depth in a muscle-equivalent liquidphantom. Notethat the power level of each waveguide Isattoo% relative to each other. The location olthe focusing probe isindicated with a red "+" sign.

10,ox-distance (cm)

(b)

-51-10

7,-----,----,---,----,

(.)

50 60 20 60

y100 SO 70 100,-100 90 60 SO

60 60 30 SO

FIGURE 14. Effect 01 array illumination on radiation pattern for the case 01 adjusted uniform illumination: (a) array illumina­tion (in percent of the selected power level of 5 W) and (b) resulting two-dimensional radiation pallern (in lOot. power levels)of the Microtherm-1000 using lincoln Laboratory adaptive·focusing algorithm with 4-cm focal depth in a muscle-equivalentliquid phantom. Note that the power level 01 each waveguide has been adjusted to provide equal amplitude as measured ata receive probe located at the 4-cm deplh. The location of the focusing probe is indicated with a red "+" sign.

10,ox-distance (cm)

(b)

-51-10

7,------,-----,----.-----,

(.)

50 60 20 60

Y100 40 40 100

:'-100 50 30 SO

60 60 30 SO

FIGURE t5. Effect of array illumination on radiation pallern for the case of inverse-tapered illumination: (a) array illumina­tion (in percent of the selected power level of 5 W) and (b) resulting two-dimensional radiation pallern (in 10% power levels)of the Mlcrotherm-tOOO using Lincoln laboratory adaptive-focusing algorithm with 4-cm focal depth in a muscle-equivalentliquid phantom. The power level 01 the outer ring 01 waveguides has been adjusted to provide twice the power of the innerring of elements as measured at a receive probe located at the 4-cm depth. The location of the locusing probe is indicatedwith a red "+" sign.

"11111 , 1U1l11t' '"3 I'lll.COlI llWlllDn .IllnUI 283

• FENN, OIEOElUCH, AND STAUfFl-;RAll Ad'lplil.~-FfKlIJillg Algorithm fur tJ Mirroll''If>r Plm"lr PllIIud-Army HYf'Cl'/hamia SF",m

3<l I I

iii

'" 20 - -•,u.E

/" I. -"••0"'-

• '- -

I I-10 • 5 I. 15

Iteration number

FIGURE 16. Measured E-fieJd focused at z =4 em as a function ofadaptive phase-focusing gradient-search iteration number for a muscle­equivalent liquid phantom. The power level at the focus increases by17 dB as a result of the Lincoln Laboratory adaptive gradient-searchalgorithm.

covering a range from 0° to 360°. We chose a maxi­mum Step size!:l.4J equal co five D/A states, which wefelt would guaramee convergence and stability of thegradient~search algorithm. Each iteration rook lessthan I min to execute; thus the array was focusedautomatically in less than 10 min, which should be ashon enough time for patient ther:tpy. We did notinvestigate varying the step size tlrp, which would haveaffeC!ed [he rate of convergence.

The results from these investigations indicate thatthe low number (16) of array antenna clements, the

high anenuation of the signal from the outer ele~

menrs in the array. and the irregular beam patterns ofrhe individual elemems prevent the Micrmherm-l 000from producing a well-defined focus in lossy muscletissue al any appreciable depth beyond 3 cm. BecausetwO proven types of phase~oplimiz..ltion routines­the systemalic ilCrativc-search algorithm and the gra~

dient-search algorithm-have yielded similar results.we are confident that the measured data truly depicllhe best possible performance of the Microtherm­

1000 system.The usc of beam shaping and different illumina­

tion smllegies to improve the penetradon depth couldbe investigated in the fUlme, but significant improve­ment is not cxpected. Instead, a better strategy for

improving the effective penelration depth might be[he use of a curved phased-array applicator with theradiating antenna e1emenu pointed directly towardthe focus {Q provide nearly equivalent path lengthsro the focal point from each radiating element. Thistype of array geometry will be investigated infuture measurements.

Conclusion

The power-deposition capabilities ofa microwave pla­Ilar phased-array hyperthermia system using an adap­tive-focusing algorithm have been characterized ex­perimentally at the University of California at SanFrancisco. The measurements shown in this articledemonstrate that an adaptive feedback gradient-searchfoclIsing compUler algorithm developed at LincolnLaboratory can be used to control the E-field radia~

tion pattern of the MicrOlherm-1 000. a commercial

hyperthermia system consisting of a planar phasedarray with 16 antennas operating at 915 MHz. In OLlrexperiments, the algorithm was used {Q focus the E­IIdd (i.e., maximi7.e the power deposition) of the

phased array at a desired target position.The adaptive algorithm uses a gradient-search feed­

back technique (method of steepest ascent). In thealgorithm, transmil-weight phase dithering is used

• FENN, DU'D1:R1CI-I. "NO ST"UFJ:ERAll Ad"pril~-FomJillg Algorithm fir fl MUnJw,/vr I'umar 1'I"md-ArrflJ HJ~rlhrrmia Splt'lII

along widl E-field probe power measuremClllS at thedesired target position to calculate the requiredgradient-search directions for sequentially and itera­tively focusing all antenna element phase shifters ofthe array. The algorithm produces t'\vo-dimensionalradiation pauerns thai arc similar 10 those producedby a systematic iterative-search algorithm that hadbeen demonsrrated previously at UCSF.

E-field focusing is intended to maximize the mi­

crowave power delivered 10 a tumor site relative 10 thepower deposiled in surrounding normal tissues. Thedata presented in this 3.rtide suggest that this goal canbe achieved for shallow tumors (i.e., tumors less than3 em beneath the skin surface) with a microwaveplanar phased-array hyperthermia system using adap­tive focusing. Measurements have proven that a pla­nar phased army at 915 MHz c.:m provide a usefulfocus in lossy muscle tissue (dielectric loss of 3 dBIem) at a maximum depth of 3 em.

With the hardware and software modificationsimplemented at UCSF, the 16-channel 915-MHzMicrotherm-IOOO system can now serve as a testbed

for new types of adaptive phased-array applicalOrs.Fumre measurementS may investigate other types ofapplicators, for example, a new noninvasive adaptivephased array of monopole antennas operating at915 MHz (recently designed and fabricated at Lin­coln Laboratory [66]). Other non-planar array micro­wave applicators arc under development that shouldprovide much needed geomerric flexibility for con­forming (Q contoured body surfaces. Conformablespiral microslTip antennas [461 have been used suc­cessfully for superficial hearing with noncoherent ar­rays, bUI (hese armys will very likely prove difficult tofocus a[ depth because of the radiated electric field,which is circularly polarized and has a somewhal

higher normal field component. An alternative designofa lightweight, conformable microwave antenna hasbeen repoTted ill Reference 67. Although this antennaalso has a complex radialed field with an electric fieldoriented radially across a circumferential gap and asignificanr normal field component. this multi-ele­ment array has demonstrated phase focusing:H depthin preliminary experiments at UCSF [681. Perhapsmost suitable for usc as a multi-element conformable

array applicator for phase-focused heating at depth is

the Current Sheet Applicator design [44,69], whichhas a linearly polarized elecrric field oriented tangen­tialtO the body surface for the simplified phasing ofadjacent elements.

Acknowledgments

The authors are grateful for their technical discus­sions with Dr. Everette C. Burdene, formerly ofL3bthermics Technologies. Inc., and currently with

Dornier Medical Systems.This work was sponsored by {he Depamnem of

[he Air Force and the N:l.[ional Insrilutes of Heallh.

REFERENCES

I. M.l. Skolnik. l"troI.!I,,'riOIl to RDdor Spurnt. 2nd 0;0. (McGraw­Hill. Nrw York, 1980).

2. R.T. Complon. Jr.. Adnpriw AmrmJaJ, Col/aplt and Prrft,..mnnrt{Pr("nlic( Hall. Englewood Cliffi. NJ. 1988).

3. A. Farina. AllltII/la-&md Sigllnl Promsi1lg TtdmiqutJ fOr RD­dnr Spltlm (Art«:h HOIIS(", Norwood. MA. 1992).

4. R. Nit1.be:rg, Adnplil1l'Sir;1a! Pr«tJ1illgfor Ratkr(Arlccll HoUS(",Norwood, MA. 1992).

S. A.J. F("Il11, "Applic:adon of A,bptive Nulling 10 Ele<:lromag­lIe(ic Hyperthermia for ImproveJ TIl(rmal nose, Dislributionin C111ccr Ther:lpy: TnlJlliral Rtporl 911, Lincoln LaOOr:llOry(3 July 1991), DTIC tAD-A24 1026.

6. A.J. Fenn. "Focuse<! N~~r.Field Nulling for Ad~p,iY( E1ecrro­ll1~gne!ie Hypenh("rmia AI>plic:alion...." /991 Pmg. Eltmom'/gt/.Rtf. SJ"'p. Prrx. Coli/brit/gr. MA. /-Sjllly /991. p. 393.

7. A.J. Fenn, MAdaplive Hyperthermia for Improved ThermalDo.'ic: DiSlribUlion.

M

in Ni,diil/iol/ RNmrrh: A TII't'1I/ltlf,..Cm/ll­'} Prrsprrtivt I: ColIgms Abrmu:ts, 1-12 July 199/, o;Os. J.D.Ch3pm~n.W.e. DL"Wq, and G.F. Whilmor'( (AC2dernic Press.San Diq;ll. 1991). p. 290.

8. A.J. Fenn, MNoninY:t:>ive Adaplive Nulling for Improvo;O Hy­perthemli:t Thermal~ DiSlribulion." 1£££ EII.ff' Mrt/irilltBiol Sot. 1m. Ctmf 13, 976 (310<:1.-3 Nov. 1'>91).

9. A.J. Fenn ant! G.A. King. MAn Ad:lpli"" Radio-FrequencyHyperthermia l'hased-Arr:lY Sy:;rem for Inlproved Canccr Ther­~py; I'h:lrllOm Targcl Masurcmenls," Trd/lliml Rrporl 999.Lincoln LaOOr:llOry (19 Nov. 1993). ESC-TR-93-2')1),

10. G.A, King and A.J. F("nn. "Adaplive Nulling MC.lSurCmClllswiLh Ihe I~SD Sigm:r-60 Allplicuor for Improved Hyperther­mia Thermal Disrribmion: Sixth 1m. eongrm 1/11 1J'~rI"rr­

",irO"ro/. J. 26Apr.-1 MIl] 1992, p. 246.II. A.J. Fcnn :lIlJ G.A. King. MAd311live Nulling in Ihe Hyper­

rhermia Tr(":llmClll of Cancer," Ullr. Lab.}. 5, 223 0'}92).12. A.J. Fenn ~nd G.A. King, "An Adaptive Radio-Frtqueney

Hyperlhemlia Ph:l.'lrt!-Arr;ly SY:;lcm for Improved Cancer Ther­apy: Phan!om Targct M(;l.';Urelllellts," '0 Ix: publl~hed in Jilt.

j. HypmlJt'mtii/.13. P.F. Turner, T. Schaefermcyer. ~nd T. Sal!lOn. MFuturc Trends

in HC:lling Technology of Decp-Se:ued Tumors," Rrrttll Rr..

~OlUM[ & _UM&[~ 7 1UJ TH! U_CO\H (UORA10RV JOUUIl 285

• FENN, DIEDERICII, AND ~AUFFf.I(

All AdtlprilJ(-PocIIJi"g Algorilbmfir /I Mirrorml'r P{mwr PbllSrd-AmlY ffyprrt/)rrmill 5pu/II

mfll Om,"~. 107,2.49 (1988).14. P.F. Turner, A. Tumeh, :lIld T. Sc:haefermeyer, -BSD-2000

APPr03Ch for Dc:.:p Loc.l and Rq;ional Hypcnherrni:l: Physicsand Tc:ehnology," StmMr/lllxrtlpir DlIJ:oIogir 165, 738 (1989).

15. G.H. Nussbaurn, "QualilY Assessment and Assurance il\ Clin_ic! Hyperthermia: Rc:quirc:menrs and Procedures,· GmrrrRf1. (51Ippl.), p. 44:481 Is (1984).

16. D.L Denman, M.J. Kolas;t, H.R. Elson, BS AWll. amIJ.G.Kereikc:s, "Specific Absorplion Ralc:.'i in Simulated TiSl\ue Me­db for a 10 x 10 cm 915-MHl Waveguide: Applic:llor,· Mrd.PI1JS. 14,681 (987).

17. CK. Chou, J.A. McDougall, K.W. Chan. and K.H. Luk,"Evalualion ofGplive Bolus Applicalors," M"J. PhyJ. 17,705(1990).

18. W.L Straube:, R.J. Myerson, B. Emami. ami L.B. Lc:yhovkh."SAR P:me:ms ofExlernal915-M Hl Microwave ApplicalOrs,"flit. j. Hyprrthrnlli/l 6, 665 (1990).

19. P. Fessenden, D.S. Kapp. E.R. Lc:.:, and T.v. Samulski, "Clin­ical Microwave Applicllor Design,· in Biufogimf. PbyJimi alldC1illiml Asp«ts uf HJpmhmnill, Mrtlital Plpirs MOl/ogmph16. c:ds. B.R. P;Jiw::l1, F.W. Henel, and M.W. Dl'Whirst (Amer·ican Institule of Physics, New York. (988). I'p. 123-131.

20. CA. Perez. B. Gillespie, T. Pajak, N. Hornback. B. Emami.and I'. Rubin, -Quality Assurance Problems in Clinka! Hy­penhermia and Irs Impact on Thernpeulic Outcome: A Re­pon by lhe R~di3don Thempy Oncology Group,~ fm. j.RttdimiOIl Dlleol. BioL PllJ'. 16, 551 (1989).

21. M.D. Shtrar, D.j. Newcomb.:, W.B. T:lylor, and J.W. Hunt,"'n"C:Slig;nion of HC'"~ling !'allenlS of {h~ MA-lOO and MA­120 MicroW:.lve Applicators undtr DilTcreru Wan,r Bolus Ar­rnngemenrs al 915 MHz,· in RJ,dillfioll Hnmrrh: A T/4¥"minb­ummy P=p«ti/Jt I: COIIgrt'1J Afmmru, 7-/2july 1991, eds.J.D. Chapman, W.e. Dewey, and G.F. Whi,more (AcademicPress, San Diq,'O, 1991), p. 194.

22. T.V. Samulski,l'. Fessenden, E.R. Lee, 0.5. Kapp. E. Tanahe,31\d A. McEuen. "Spiral Micro""rip Hn>cnhermia Applica­10l"!i: Technical Design :lIld Clinicall'erformance:,~ fm./ Rmli­tl/ioll Ollrol. BioI. Ph,S. 18, 233 (I 990).

2}. A. Tcnnalll.), Conway, illld A.I'. Anderson, "A Rolxlt-Con­lrolled Microw.l.ve Antenna S»,em for Uniform Hyperrher­mia Tre-JUllelU of Superficial Tunult"li with Arhitrary Shape,~

Im./ Hyprrrbmlli/l6, 193 (1990).24. "Spc:ebllssue on PhaStd Army5 for Hypenhemlia TrC:lImenr

ofC"ncer." fEEETMlu. Mimm •. 71m}/] Trrh. MIT-34 (May1986).

25. "Special lSl\ue on Hyperthermia and c',nlXr Therapy." lEEETTl/III. Biomm. Eng. BME-3/ (J:tn. 198<1}.

26. V. Salhi:Lscelan, M.F, Isklllder, G.C.W, Howard, and N.M.B!ee!ltn, "Theoretical Allalysb alld Clinical Dc:t1lon~lr.llion ofthe Effect of I'ower Control Using rill: Annllbr Pha.',ed-ArrJyHypenhermia S~lem: IEEE Tm/IJ. Mirro",. 71Jrory TrrIJ.""n··34, 51<1 (198(,).

27. V. SaThiascebn, "Potemial for l'atielll.SpecirlC OpTillliz;uionof Dt.'ep HC'"Jling l':lllerns through M:mipuhlion or Ampli­,mle :lIId l'hase: 5tmblmthrTi/pir DIIJ:ologir 165, 743 (1989).

28. D. Sullivan, "Thrl"C-Dimemional CompUTer SimulJtion inO«p Region;tl Hypt'r1bl'lrnb Using rhe FiniTe-DifferenceTime-Domain Mcthod: I££E TrJIIIJ. Mirror!!. Tbrory TrekMlT-38, 204 (1990).

29. B.S. Trembly. A.H. Wilson, M.j. Sl1l1ivan, A.D. S,dn, T.Z.Wong, and J.W. S'rohhehn, "Conlrol of fhe SAR I'Jlle:tnwilhin an Imem\lial MicrowJve Array lhrough VJrialion ofAnfcnnJ Driving ['h;=," fEEE Tmm. Mirrow. TIND'] TrrlJ.MlT-34, 568 (1986).

286 lHllt~COI. IlBORI10RY jDUUIL fOLUWI~ .•UWBI~ 7 t~91

30. G. SalO. C ShibJta. S, &kimubi, H. Wak;lbJ~·Jshi. K. Miu:u­ka, and K. Giga. "Phase-Controlled Circular Arr.t)' HC':lringElluipmem ror Deep-SeJled Tumors: l'reliminary Experi­menu:: IEEE TrflllJ. Mirrow. Throry Trrll. MlT·.H. 520(1986).

31. I'A. Clldd. A.I'. Anderson. }"tS. Hawlc:>" and J. Conway,"pnas<.'(j-Arr.lY lXsign Col1sidcr.l1ions for D«p Hype:nhc:rmiaIhrough Layerro TissTle," IEEE T(l/1/$. MlnTll/'. T/,..o,] Treh.MIT-34, 526 (1986).

32. N. Morita, T. Hamas;,ki, and N. KumagJi. "An OplimalExci,alion Melhod in Mulri:tpplic:uor SYSTetl\5 for Forming aHOI Zone in:;ide lhe Human Body," IEEE Tr./JIJ. Mirrow.Throry Trrh. MTf-34, 532 (1986).

33. e. De Wagler. "Opdrnit:llion ofSitnubcc:d Two-Dimcr15ion­al Temper:nure DiSHiburions Induced by Muldple Electt\)­magnetic Applicalors," IEEE Tral/s. Mimuli. T/,..ory TrekMTf-34, 589 (1986).

34. M. Knudsen and U. Hartmann, "Optimal T cmpernture Con­lrol with Phased-Array Hypenhe:rmia Syslem: IEEE Trill/$.MinTIlIJ. Throry Tt'flJ. MTf-34, 597 (1986).

35. CF. Babb5. VA Vaguine, and J.T. Jones, "A Prcdicrive­Ach!plive:, Muhipoim Ft'edb3ck Conrroller for Loc:tl Ht:ll Ther­apy of Solid Tltmors.~ IEEE Tram. MinTI/li. ThrrJry Trrh.MTf-34,604(1986).

36. R.B. Roemer, "Physical and Enginl'l:ring AspeclS of H}'per­lhermi~," !'roc, 8th fm. Gmgrm &dimiofl Rrs. 2, jllly 1987,eds. E.M. Fielden, J.F. Fowler, J.H. Hendry, and D.O. SCOll,

p.948.37. R.B. Roxlller, K. Hynynen, C.)ohnson, and R. Kress, "F~­

b.,ck Comrol and OpTimi1.alioll of Hyperthermia Hc:lling!'atterlU: ]lm;c:1\l SI:tIUS and Fumre Ncffis," IEEE EigIJtlJ An­/111111 Coli! E"g. Mrd. Biol5«.. NOli. 1986. p. 1496.

38. R.B, Roemer, "Optimal Power Deposilion in Hyper­thermia. I. The Trealmcm Goal: The ldc:ll Tcrnpc:raTurc:OiwihuTion: The Rolc: of Large Blood Vesscls," fm. j.HyprrrhrnlliJ/7, 317 (1991).

39. t\. Bo,lg and Y. Levl,llan, "Opllmal Ex,iwfi"n of Muhi~p,.lj­calOr SYSlClllS ror D«p Rcgional Hypcrthc:rmi:t." lEEE Tram.Biomrd. Ellg. BME-37. 987 (]990).

40. I'. WUSI,J. Natlohny, R. Felix, P. Ocuflh:lrd, A. l.ouis, ;lIld W.John. ~Slratcgio ror Opli mite,l Al'l'liC:llion orAtlntrlar-I'h:J.«c:d­Ar..~y S~terns in Clinical Hn,.,rthC:rlnia." 1m. J. HJprrrlJrmJil17,157 (1991).

41. OJ\. Nguyell amlJ.W. Slrohbc:hn, "Oplirnall'osilions, Ori·erll:1Iiol1s, Amplittrdc::s and Phase:; of Movable Microwave AI'­I'licalOrs in Hyperthe:rmia T realmenl: in HJprrrfxrmic DlJroL1988 I, nis. 1'. Sug.,h3ra and M. SailO rraylor & Francis,London, (989), pp. 738-739.

42. ).W. Struhbdm, E.H. Cunis, K.D. I':lulsen, and D.R. Lynch."Optirnil':llion of lhe Absorbed Power Dislrlbulion for anAnnular l'ha......J-Arr.TY Hypenhcrmia Syslem: 1m. j. Nt,dill­thlll Duro!. !Jiol. f'llJS. 16,589 (1989).

43. J.T. !..o.,ne III and S.\'(I. Lee, ~G~in Optirni7.:uion of a NC3r­Field Focusing Array for Hyperthermia Apl)licalil)n~,~ fEEETrtltJJ. Murow. Throry Trrk M1T·37, 1629 (1989).

44. M.K. G0l'al, J.\'(I, Hand, M.L.O. LlIlllori. S, Alkhairi. K.D.Paulsen, :Uld T.e. c.:la.~, "Current Shcrt AppliC'.lIOr Arrays forSuperficial HYI".:rthermia ofChol Wall Loion~: II/(. / Hy­prrIIJrrm;'t8. 227 (1992).

45. H.R. UoderworK!, A.F. l'et<'i"S<ll1, amI RL M~gin, ~EJeclri,­

Field DisTribltlion NC"Jr Reclangular Microstrip RadiaTOrs forH}"perthcrmia Hc:lling: Theory ve:nus ExpcrilllC:1J( in \'Valc:r,~

IEEE TmflS. Biomrd. Ellg. 39, 146 (1992).46. E.R. Lee, T.R. Wilsc:y, P. Taray-Hornoch, OS. Kapp, P.

.. f ..• FF...'\,IN, DIEDERJCH, AND STAUFFER

All Adnptil...-Focm;ng Algorirhm fo, II Mirrouwll<' PlllIUlr I'JJllScd-AmlJ Hyptrlbtrmia Spt,,,,

FCS5Cnden, A. l.ohrh:lch, and S. [)riona.s, ~Body Confornuble915 MHz Micromip Arr:lY Applic;llors for Large SurF..cc AreaHYlXnhermia,· IEEE Tram. Biomrd Ellg. 39, 470 (1992).

47. 0.$. Karp, P. Fessenden. T.v. Saffiulski. MA. Bagshaw, R.S.Cox, E.R. lei:, A.W. Lohrbach.J.L Mqer, and S.D. Prionas,"Stanford Univ..rsity InSliludonal Repon. I'h:lSC I Ev:UuationofEquipmcm for HYlX"Tlhcrmia T,calment orCanecr, H /",.;.

Hypmhmnia4. 75 (1988).48. J.W. Hand, J.L eha-tham. and A.J. /-lind. M AbsorOOI Power

Distributions from Cohcwu Microw:l.VC Atr.I.)IS for l.oa.lizcdH)'pen:hcrmia.~ IE££ Tm'u. Mirrow. Tixory T«h. MIT·)4,484 (1986).

·19. c.]. Dia:lerich and P.R. S[;lulTct, -Prr-Oinia.! E\'2J~lionofaMicrowave Pbnar Arr;ty Applicator for Superficial Hypcnhcr­mia.~ 11If.). HyfJ"TlKrnTia9, 227(1993).

50. e.). Dic(krich. G. Sherwin, e. Adams, amI P.R. Stauffer,"Evaluation of a Multi·E1ement Microw:lve Applicilor forHyf"C'rthermia: in RRdilltiolr RtsMrr:b: A 7iwlllittlrCmttlryI>mpmi~ I: Cot/gm! Almrnrts. 7-12 jllly 1991, als. J.D.Chapma'l. W.e. Dewey, and G.F. Whitmore{AcademicPress,San Diego. 1991), p. 194.

,SI. H.!. Bassen, M. Swicord, and). /\bit3,"A Mini;l{utc Broad­Band Eleenic Fidd Probe," Annals ofthr Nnv York Acndt",y ofScimm. Biologiml Effim ofNrmioniung RAdimioll 247, 48 [(1975).

52. H.I. Bassc::n and GS. Smith. ~E1eetric Field Probcs-A Re­\'ir-\'," IEEE Tnws. Allin/lUIS Propllg. Ap·)I, 71 0 (S<:pt. 1983).

53. A. Uhlir, ~Char.Kteri7.:ltion ofC~tal Diodes for low·LevelMicrow;we Deleclion," MicTOII'.;. 6, 59 (July 1%3).

54. CH. Durney ;Ind M.F. [skander, ~Amenn:lS for MediCl1/\p­plications," in Allfnll/II Hrwdbook: TIND? Applir'(l/iom, 'IlldDrrign. cds. Y.T. lo and S.W. Lee: (Vall N05lr.1l1d Reinhold,New York. 1988).

55. R.L Zahr:ldnik, TJltOry lind Trrlmi'llits ofOptimiZlltion forPm(tid/lg £I/gi/lt'l'f"l (13."ncs :lIld Nob)e, New York, 1971), pp.118-124.

56. L HasdorlT. CmdirlU OptimiZluioli mul NOlllhltilr Comro/(John Wiley, New York, (976).

57. D.). Far;na and R.I'. Flam, "A Stlf-Normalizing Gr:tdiem_Sc:udt Adaptiw: Array Algorithm," 1£££ TmllS. ArroJp. £/u'­Jrol/. Eftt_Z7, 'JOl (19'11).

58. A. C.1nlOni. MApp.Jication ofOrthogonal Perturbation ~uenccsto Adapti\'c Ikamformi"g,~ IEEE Tmm. AlltrmlflS Prop"g.Ap-28, 19J (1980).

59. LC Godara and A. GnlOni, "Analysis of the Performance ofI\dal'tivc Ik:Jm Forming Using ['crturb.,tion Sequences: 11:.""££T,.,ml. Al/lrmi<ll Prop"g. Ap·}I, 268 (198}).

GO. D.C Fardcn and R.M. Davis, "Onhogonal Weighl Perturba­lion Algorithms in Panblly Adaptive Arrays: IE££ Trmu.Altti.'1111111 PropnG' AP-}}, 56 (1'185).

61. A.j.l'enn, "Maximizing Jammer Effectiveness fOt E....~lu:ttinglht" l'erfonllanee of Adaptive Nulling Arr:lY Amenn:lS." IE£E]'ml/S. Amr/1I/1I1 I'rop',g. AI'·33, 113 J (Oct. [985).

62. A,). Fen", e.). Diederich, and P.R. St:mffcr. "ExpcrilllclllalE\':tluatioll of all Ad:tplive Focusing Algorithm fOt a Micro­wave ['b"ar I'h:tScd.Array Hyf"C'nhermia SyMcm at UCSF:Trd",mil Rrpon 977. Lincoln labor.nory (17 May 1993),ESC-TR·92·203.

63. T.V. Samulski :lnd E.R. Lee:, "Pn:diniol Ev:J.lu:ttion of Lux­Imn 3000 SptCIl1," IEEE Eighth A/lI",nl CO/if. £"g. Mrd. Biol5«., Fort \'(Irmh. ]X, 7-/0 Nov. 1986. I). 1493.

64. G. H:msgro\'e, A. Kr:ISl.cwski. :Ind A. Surowiec, "SimubtcJBiologic:tl M:tteri:lls for Elecnomagnetic Radiation Absorp.lion Sludia, Bio/orlrrtrollillg>mics 8, 29 ([987).

6,S. J.W. Hand. "Biophysics and T cchnology of ElectromagneticHyperthermia," in MrthodJ of&rmu,1 HyprrrlKrnTi< Hrating.ed. M. G:..utheJie (Springer-Verlag. Berlin, 1990), pp. 1-59.

66. A.J. Fenn, O.S. roc, CE. Reuler, :lnd A. Taflove, -Noninva­sive Monopole Ph;ISCd ArI1l.Y for Hypenhamia Treatment ofCl1I.nial-uvity and Skull·B;ISC T umOts: Daig", Analysis, andPhantom T CS15," IEEE Am/lUll Con! Eng. Mtti.. Bioi 5«.. SailDirgo. 0i 15, part 3. p. 1453 (1993).

67. P.R. Stauffer, C}. Diederich, :and D. Bozzo, "lUdial E·FiddMict'OW:lve Antenna for Superficial Hyperthermia via Singleor Multi·Element Conformal Apertures." Sixth Int, Congrm ofHypmhmnit Ontoloo I, cd. E.W. Gerner. p. 319 (1992).

68. P.R. Siauffer:and C). Diedcrich, M915 MHz Conformal Am.yMicrow;a\'c Applicllor." Prot. Nonh Ammtlln Hypmhcmia5«. Mrg.• Dallas, rx p. 13 (1993).

69. M.K. Gopal ;lnd T.e. Cctas, ~Current Shcct Applic:nors forOinical Micll)\'.'3''e Hyperthermia; IEEE Trans. Mi(TtJw. TJ1t­II? Ttth. MlT·41, 431 (1993).

YOtUll1 6 NUW81~ 2 U!lJ tHE LlNC01NlllORUORI JOURNIL 287

• FENN. DIEDERICH. ANI) STAUFFERAll Adnptil~Fo{l4SlI/X Algorilhm ftr (/ MirrowlIl'r Plal/Ilr PJmsrd-ATTIIJ Hyprlb..rmill SJSI<"III

..

ALAN J. FliNN

received the following dc:grtt.'iin eleclnal engineering: a B.S.in 1974 from the University oflJJinois in Chiago and all M.S.and Ph.D. in 1976 and 1978,rcspteli ...ely. from Ohio Sr:neVniversicy in Columbus.

From 1974 to 1978, Alan~ a graduate r=rch associ­ate al the Ohio State: Vniversicy8ecuoScienee Labor-nory,whe:re: he: did rc:sc::lrc:h inphasc:d-arr.ty antenna analysis.He later beame: a senior engi­nc:<:r at Manin Mariett;l Aero­space in Denver. where: he~involved in br~doond antennarc:scarc:h and developmem from1978 to 1981. He joined theSLafT of Lincoln labor-nory in1981 and~ a member of the:Space Radar Technology Groupfrom 1982 to 1991, where hisrcscarc:h was in phased-atr.lYantenna dcsign, adaptive-arraynear-field tcsling, and anu:nnaand r:ldar el'O$S-scetion mea­sure:ments. Since 1990 he hasbeen eonduCling rest;:lrc:h in theapplication of adaptive-nullingtechniques to radio frequency(RF) hype:nhermia tre;ltnlenl.He is currently an =istantleader in the RF TechnologyGroup, where he is investigat­ing optoelectronic imegr:lted-

circuit phased-arr.ty amennasfor mobile S:llellilt communica­lions applications.

In 1990 Alan w.IS acorceipitm of the IEEE Ante:n­nas & Propagation Sodtly'sH.A. Whc:<:Ie:r ApplicationsPrizt raptr Award for "l'h.lS<'dArray Antenna Calibr:ltion andPautrn Prtdielion Using Mu­tual Coupling Mtasure:mtnts,'a paptr Ihat he: coauthored forthe 1£££ Trr1>lSilrriMS 1)1/ Amm­1JJlS mull'ropng,ll;Q1/. Ht Sl'rvedas an associate: editor in tht are:aof adaptive arrays for the: 1£££1h11lSIUl;Q'/S fin AlIfrl1>las andPropaglltio" from 1989 to

1991. He is a senior member ofthe IEEE and a member of du::North American HyperthermiaSociety. and has bttn appoimcJ{O a five-year t('fm of member­ship in the: Institute for Systenuand Components of the8ec,romagneda ACldcmy.

He is the author or co­aUlhor of more Ihan 20 articlesin his fidd and has presentedmore than 20 P.1pers al nationaland international societymc:<:tings.

CIIRIS J. DI£OERtCtl

received a B.S. degree inbioeuginccring from the Uni­versity of California, SanDjll,'O, in 1984, and an M.S.and a Ph.D. degree in electricalengineering from the Vnivcr­siryofAriwnain 1986 and1990, resp«:livdy. Ht is an;!.'\Sistant adjunct profo:ssor atlhe UniversityofDJifornia,San Fr.tneisco, RadiationOncology Department, where:his focus ofrc:serrch has Ix-tnon the design and developmentofhypenhermia devices andother thermal therapy ddiVl:ryS)'Slcms. Chris is a member ofthe Nonh Americlll Hyper­thermia Socicry.

PAUl R. STAUI'I'I'.R

n~ctived a BoA. <Iegrcc in phys­ics from the College ofWOOSterin 1975 and an M.S. dq;rc:<: indectrial cngineering from theVni...el$ity ofAriwna in 1979.He is an :tSSOCiate adjunctprofessor at the University ofC:llifornia, San Francisco,Radiation Oncology Depan~

menl. where his focus of r('~

search has been on thc engi­neering dcvelopmellt of RF,microw:we, and uhl':lSOundhyperthermia technologies. Hehas receivN a clinical engincc:r­ingcerlification from thellllernaliollal CertificationCommission and a mediClIphysies cerrificatinn in hyper­thermia from the Amer;c:tnBo.1rd of MediClll'hysia. He isa mcmbe:rof the North Am('ri­e:tn Hypenhennia Society.

288 rH( lINeOL' t.llO~1101Y JOU~~.t ~lltUIII 6. ~UIlII'11U3