Radiation patterns of dual concentric conductor microstrip antennas for superficial hyperthermia

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 45, NO. 5, MAY 1998 605

Radiation Patterns of Dual Concentric ConductorMicrostrip Antennas for Superficial Hyperthermia

Paul R. Stauffer,* Francesca Rossetto, Marco Leoncini, and Guido Biffi Gentilli

Abstract—The finite difference time domain (FDTD) methodhas been used to calculate electromagnetic radiation patternsfrom 915-MHz dual concentric conductor (DCC) microwaveantennas that are constructed from thin and flexible printedcircuit board (PCB) materials. Radiated field distributions arecalculated in homogeneous lossy muscle tissue loads located undervariable thickness coupling bolus layers. This effort extendsthe results of previous investigations to consider more realisticapplicator configurations with smaller 2-cm-square apertures anddifferent coupling bolus materials and thicknesses, as well asvarious spacings of multiple-element arrays. Results are givenfor practical applicator designs with microstrip feedlines etchedon the backside of the PCB antenna array instead of previouslytested bulky coaxial-cable feedline connections to each radiatingaperture. The results demonstrate that for an optimum couplingbolus thickness of 2.5–5 mm, the thin, flexible, and lightweightDCC antennas produce effective heating to the periphery ofeach aperture to a depth of approximately 1 cm, and may becombined into arrays for uniform heating of large area superficialtissue regions with the 50% power deposition contour conformingclosely to the outer perimeter of the array.

Index Terms—Array applicator, FDTD modeling, flexible an-tennas, hyperthermia, microwave antennas, power deposition,tissue heating.

I. INTRODUCTION

L OCAL hyperthermia has been shown to be an effectiveanticancer treatment when used as an adjuvant to radia-

tion and chemotherapy at total thermal doses above about 100cumulative equivalent min at 43C (CEM43) or 400 min at42 C [1]–[4]. Breast cancer is one of the most prevalent formsof cancer in women and, unfortunately, local control remainsa problem of increasing magnitude. Chestwall recurrence ofbreast carcinoma is a common problem that is resistant tostandard therapies and often spreads to large diffuse areas onthe chestwall, abdomen and back—extending from the skinsurface to a depth of 0.5–1.5 cm. While randomized clinicaltrials have clearly demonstrated the efficacy of adjuvant hy-perthermia in smaller lesions that can be accommodated withexisting equipment [5]–[7], the number of patients that canbe heated adequately with current single and multiaperture

Manuscript received June 6, 1997; revised November 19, 1997. This workwas supported by the National Institute of Health under Grants RO1 CA-70761 and R43 CA-69868.Asterisk indicates corresponding author.

*P. R. Stauffer is with the Department of Radiation Oncology, University ofCalifornia, 505 Parnassus Avenue, P.O. Box 0226, San Francisco, CA 94143-0226 USA (e-mail: [email protected]).

F. Rossetto is with the Department of Radiation Oncology, University ofCalifornia, San Francisco, CA 94143-0226 USA

M. Loencini and G. Biffi Gentilli are with the Department of ElectricalEngineering at the University of Florence, Florence, Italy.

Publisher Item Identifier S 0018-9294(98)02878-X.

planar array applicators is severely restricted. Thus there isa real need for flexible applicators that can uniformly heatlarge area superficial disease overlying complex contouredanatomy. Microwave radiation at the ISM band frequencyof 915 MHz has a half power penetration depth in humansoft tissue of about 1–1.5 cm [8], [9]. Thus, it appears wellsuited for treatment of superficial chestwall disease if appliedvia a sufficiently large applicator with temperature-regulatedcoupling bolus to control the skin surface temperature andsmooth power deposition (SAR) peaks under the apertures.

Previous efforts of our group have characterized the powerdeposition patterns, or specific absorption rate (SAR) profiles,of single and multiaperture microwave applicators constructedfrom 915-MHz dual concentric conductor (DCC) radiators ofseveral sizes [10], [11]. Prior work has also demonstrated theclose correlation of finite difference time domain (FDTD)-based theoretical simulations like those presented in this paperwith measured SAR patterns for 3.5-cm-square DCC aper-tures [10]–[12]. In nearly all these previous studies, radiationpatterns were calculated assuming an ideal equal-amplitude,equal-phase sinusoidal field excitation across all four cornersof the square DCC apertures, a configuration that is simple tomodel, but awkward to realize physically.

The current effort describes the radiation patterns obtainedin homogeneous lossy muscle tissue from square DCC apertureantennas excited via microstrip feedlines located on the backsurface of thin and flexible double-sided printed circuit board(PCB) material. To demonstrate the capability of higher reso-lution control over the heat distribution than has been possiblewith previous microwave applicators, a smaller 2-cm-squareDCC aperture is investigated in this study. The effectivenessof different coupling bolus layers is compared by examiningthe SAR in vertical plane (depth) cross sections and in planesparallel to the surface and either 0.5 or 1-cm deep in muscle.Finally, the heating uniformity under a six-element array isanalyzed to demonstrate the capability of building arbitrarilylarge applicators with the potential treatment area limited onlyby the number of available microwave amplifiers and not byfixed applicator dimensions.

II. M ETHODS

A. Applicator and Load Configuration

The present effort concerns the theoretical analysis of ra-diation patterns from square DCC apertures etched on thefront surface of flexible double-sided copper PCB material andexcited at a fixed frequency of 915 MHz. Radiation patterns

0018–9294/98$10.00 1998 IEEE

606 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 45, NO. 5, MAY 1998

(a)

(b)

Fig. 1. Basic FDTD model geometry showing square DCC slot aperture infront surface copper ground plane, microstrip feedline on back surface of thinPCB material, and layered dielectric load with absorbing boundaries.

were calculated in homogeneous muscle tissue ( 56,1.4 S/m) in front of a single aperture DCC antenna formedby a 1.8-cm-square copper patch surrounded by a 1-mm-wideslot of side length 2 cm (Fig. 1), and for six-element arrays of2-cm-square apertures (Fig. 2). Analysis of the single-elementconfiguration was performed simulating an infinite layer ofmetallization around the square slot by extending the groundedmetal layer along the whole calculation domain which included3 cm of muscle around the aperture and an additional 4 cmof absorbing boundary layers. For the six-element array, threeconfigurations were studied with apertures always spaced 0.75cm apart horizontally, and either 0.75, 1.25, or 1.5 cm apartvertically. Instead of an infinite ground plane, a 1.5-cm-widerim of grounded metallization was used around the arrayperimeter to simulate a practical array applicator (Fig. 2).

Several different load configurations were considered for thesingle-element applicator: direct contact of the applicator withmuscle, and use of various different thickness coupling boluslayers consisting of either distilled water

Fig. 2. Configuration of array applicator showing six DCC apertures in thefront surface ground plane and microstrip feedline network on PCB backplane.

or silicone oil1 . For both couplingmaterials, bolus layers of 2.5, 5, 7.5 mm; 1- and 2-cm thicknesswere simulated. The real microstrip fed applicator [Fig. 1(b)]was modeled as a 0.254-mm (10 mil)-thick dielectric substrate2

with and and coupled to the muscle load witha 5-mm-thick water bolus. The six-element array (Fig. 2) wascharacterized with a 5-mm-thick layer of silicone oil or waterbolus between the applicator and muscle load.

B. Electromagnetic Feed Configurations

For the single-element applicator, two electromagnetic feedconfigurations that have been studied in previous experimentalwork were modeled with the FDTD analysis. First, ideal 915-MHz sinusoidal fields were enforced across the slot in thepositions shown in Fig. 3(a) and (b) to simulate excitation ofthe aperture symmetrically across all four corners or crossingjust one side gap, respectively. In addition, the real case of mi-crostrip feedline excitation [with field applied vertically acrossthe dielectric substrate between microstrip and groundplane asshown in Fig. 3(c)] was simulated for comparison with theideal one side feed [Fig. 3(b)] to verify accuracy of the simplerideal source model. In this case, a microstrip width of 0.8636mm was used to maintain a matched 50-feedline impedanceon the 0.254-mm-thick substrate material. After crossing theslot at the middle of one side, the microstrip was terminatedin an open circuit immediately behind the center of the frontsurface patch. For the array configuration, the apertures wereexcited by ideal noncoherent 915 MHz sources representingmicrostrip feedlines crossing the center of each directionslot as shown in Fig. 2.

C. FDTD Numerical Analysis

All microwave applicator radiation patterns were analyzedusing the FDTD numerical method, which easily modelsthe multilayered lossy tissue loads typical of hyperthermiaapplications. The method consists of the numerical solutionof Maxwell’s equations in the time domain using a finite

1DC-200 Fluid 50 cs available from Dow Corning Corp., Midland, MI.2Cuflon CFA-10-7-7 available from Crane Polyflon Co., Norwalk, CT.

STAUFFERet al.: RADIATION PATTERNS OF DUAL CONCENTRIC CONDUCTOR MICROSTRIP ANTENNAS 607

(a) (b)

(c)

Fig. 3. EM field excitation configurations used for the FDTD simulations (a)Ideal symmetric feed across all four corners with equal amplitude and phasesignals. (b) Ideal excitation across one side gap (�x direction). (c) Realisticexcitation of aperture from microstrip feedline crossing�x direction side gap.Ideal sources consist of a 915-MHz electromagnetic (EM) field impresseddirectly across the gap in the front surface metal layer. The microstrip feedexcitation was simulated by impressing the field between the front surfaceground plane and back surface feedline port located 1.85 cm off to one sideof the aperture.

difference approximation of the temporal and spatial partialderivatives [13]. A nonuniform grid size was used in orderto model the fields in the vicinity of the narrow gap and mi-crostrip structures accurately without unnecessarily increasingthe total number of cells representing the structure. In the-

plane a step size of 2.5 mm, representing approximatelyin muscle at 915 MHz, was used for the majority of the

structure, while finer grids of 0.5 or 0.4318 mm were usedin the regions of the gap and feedline respectively. In thevertical direction, 2.5-mm cells were used for most of themodel except for the coupling bolus and tissue interface regionwhere a finer grid to 0.125 mm was used, and in the air regionbehind the applicator where 10-mm cells were sufficient. inorder to improve accuracy of the solution for the discretizedMaxwell’s equations in the interface region near each shift ingrid size, a gradual logarithmically changing grid was used.A time step 6.4288 10 was used, correspondingto 1700 steps for each sinusoidal wave period at 915 MHz,in order to satisfy the Courant’s stability criterion [14]. Todisplay the field calculation results, the orthogonal, ,and magnitude components calculated on the sides of eachunit cell were translated to the center of each correspondingcell by interpolation. A steady-state analysis was performedusing 915-MHz sinusoidal field excitation enforced across thecenter of one side of the gap, or coherently across all fourcorners, as diagrammed in Fig. 3. In addition, a real 50-mi-

crostrip feedline was simulated, placing the driving field acrossan open-ended microstrip “port” as shown in Fig. 3(c)—adistance of 1.85 cm from the gap. To avoid reflections atthe perimeter of the model, field propagation beyond themodel boundary was simulated using four absorbing layers ofprogressively increasing conductivity as suggested by Taflove[15]. With four cells width per layer, this required a totalabsorbing boundary region of 4 cm in muscle laterally aroundthe model and about 12 cm in air above the applicator. Noadditional absorbing boundary layers were needed in thedirection due to high losses in the 3-cm-thick muscle tissueload. Due to symmetry in the model, only half the structurewas simulated. Thus for the 2-cm-square DCC aperture with 2-cm-thick bolus layer, the model had a total calculation space of16 8.1 17 cm with 294 000 cells. For all cases consideredin this study, power deposition was calculated from

where is the electrical conductivity and the density ofthe muscle tissue load. In order to correctly simulate radia-tion from a noncoherent six-element array, power deposition

was calculated individually for each of the two leftcorner elements and the two central elements when the otherfive were turned off. Fields for the two right corner elementswere derived by reflecting the calculated patterns of the lefttwo corner elements due to symmetry of the array. Finally,the six separate SAR distributions were combined accordingto the relation

Each individual simulation was carried out with the ap-propriate model geometry including adjacent aperture slots,surrounding 1.5-cm grounded rim, absorbing boundary layers,and layered tissue load. Thus, the total calculation space for thelargest array with 1.5-cm vertical spacing, 0.75-cm horizontalspacing and 0.5-cm bolus was 18.516.5 15.5 cm and thetotal number of cells was 742 000.

III. RESULTS

Fig. 4 shows the SAR distribution obtained in a horizontalplane 1-cm deep in muscle normalized to the maximumSAR value in that plane, as well as the three separate-field squared components with each one normalized to themaximum total -field squared in that plane. Thesimulations assume the applicator is fed symmetrically acrossall four corners of the aperture gap as shown in Fig. 3(a), andis coupled directly to the muscle tissue. The outer perimeterof the radiating aperture is shown to scale on each contourplot as a dashed square. In this case, the symmetry of theelectromagnetic feed excitation is reflected in the symmetryof radiated field components. Differing only by a 90rotationabout the axis, the and field components exhibit identicalshape and magnitude with peaks of about 45% of the total fieldappearing under the two opposing gaps and with very littlefield ( 9% centrally. In contrast, the component


Fig. 4. SAR andE2-field contour lines for the ideal four corner feed DCCaperture in a horizontal plane 1-cm deep in muscle when the applicatoris directly coupled to muscle. (a) SAR/SARmax, (b) E2

x=E2

TOTmax, (c)

E2y=E2

TOTmax, and (d)E2

z=E2

TOTmax. Projections of the aperture perime-

ter and radiating slot are shown as dashed lines with proper scale and position.Note the symmetry ofx andy field components and total SAR obtained withsymmetric (four corner arrows) source excitation.

field presents a single high central peak reaching a maximumof about 85% . For this symmetrically fed aperture, thecomplete SAR pattern is characterized by a centrally locatedsingle symmetric peak whose 50% of maximum SAR contourline extends out past the perimeter of the aperture except forapproximately a 2-mm portion of each corner.

For the easiest to build feedline configuration consistingof a single microstrip crossing just one of the aperture gaps[Fig. 1(a)], the SAR distribution becomes notably asymmetricwhen the applicator is coupled directly to lossy tissue, asshown in Fig. 5. In this case, the normalized SAR in the 1-cm-deep plane has a peak concentrated under the microstrip fedgap, such that the region greater than 50% of maximum SARcovers only part of the aperture region and is instead skewedoutside the aperture boundary toward the direction. Thisis due primarily to the distorted field component whichchanges to a single peak of approximately 85% oflocated directly under the microstrip fed gap. In fact, thetwo symmetric peaks of the component and the centralcomponent peak are less than 15% and 20%, respectively, ofthe total field. While such an asymmetry and skewing ofthe field off to one side of the aperture is undesirable, thesituation improves significantly when a dielectric “couplingbolus” layer is added between the applicator and muscletissue load. Fig. 6 shows the normalized-field squaredcomponents and SAR obtained in a plane 1-cm deep in musclewhen the applicator is again fed across the directiongap, but coupled to the muscle load with a 5-mm-thicksilicone oil bolus. Note that the -field components arenot completely symmetric but the distortion is significantly

Fig. 5. SAR andE2-field contour lines for the ideal�x side gap feeding ofa single aperture in a horizontal plane 1-cm deep in muscle when the appli-cator is directly coupled to muscle. (a) SAR/SARmax, (b) E2

x=E2

TOTmax,

(c) E2y=E2

TOTmax, and (d)E2

z=E2

TOTmax. Note the overwhelmingE2

x

field concentrated under the microstrip fed(�x) gap producing significantasymmetry of the total SAR.

Fig. 6. SAR andE2-field contour lines for the ideal�x side gap feedingof a single aperture in a horizontal plane 1-cm deep in muscle when a5-mm-thick coupling bolus of silicone oil is inserted between applicator andmuscle: (a) SAR/SARmax, (b) E2

x=E2

TOTmax, (c) E2

y=E2

TOTmax, and (d)

E2z=E2

TOTmax. Note the rebalancing and recentering ofE-field components

producing an SAR that is well localized under the aperture compared to Fig. 5.


(a) (b)

(c) (d)

(e) (f)

Fig. 7. SAR contour lines for a single aperture fed ideally across the�x side gap. The SAR is shown in a vertical cross section through the�x feedpoint forbolus thicknesses of (a) 0, (b) 2.5 mm, (c) 5 mm, (d) 7.5 mm, (e) 1 cm, and (f) 2 cm. SAR is normalized to its peak value in muscle just under the gap and bolus.

reduced and the and components are considerably morebalanced compared to the direct-contact aperture. Thoughstill asymmetric and connected, the component has twoidentifiable peaks reaching a maximum of 55% of the totalfield. The component contributes two symmetric peaks ofabout 30% of the total field and the component has asingle central peak with a maximum of about 40% of .The combination of these more balanced fields determines an

SAR distribution with considerably improved uniformity andlocalization under the aperture. For this configuration, the 50%SAR contour in the 1-cm-deep plane covers the entire aperturefootprint, and extends an additional 3–5 mm past the apertureboundary in both and directions parallel to the feedlineexcitation.

Fig. 7 shows the normalized SAR in a vertical plane crosssection through the middle of the aperture along thedirec-


tion for several thicknesses of dielectric coupling bolus from 0to 2.0 cm. It is readily apparent that even the minimum thick-ness 2.5-mm silicone oil layer provides significant smoothing,centering, and an increase in penetration of SAR under theaperture. The effect of progressively thicker bolus layersis to further widen the power deposition pattern and beginshifting the pattern back toward the microstrip fed side.Considering that thermal conduction and convection effects inmuscle will tend to smooth and center the resulting tissuetemperature distribution under the aperture even more, the2.5 and 5-mm-thick silicone oil layers appear nearly ideal forthis 2-cm-square aperture since thicker layers over emphasizeheating centrally as compared to the periphery, and extend theradiated fields too far outside the aperture perimeter where theycan interact unpredictably with fields from adjacent apertures.

All the above simulations were performed using ideal ex-citation across one side gap as shown in Fig. 3(a) and (b). InFig. 8, the SAR distribution obtained in the 1-cm-deep planefor ideal one side feeding is compared to that obtained fromthe same aperture excited by a real microstrip feedline, suchas that shown in Fig. 1. The direct correlation of SAR patternsdemonstrates that the ideal model of the feedline, whichrequires less than half the number of cells to characterize thestructure than the real feedline case, is accurate and appropriatefor this analysis.

Fig. 9 shows the normalized SAR distribution in a horizon-tal plane 1-cm deep in muscle for a six-element array coupledwith a 5-mm-thick silicone oil layer, when the apertures are fednoncoherently by ideally modeled microstrip feedlines fromthe direction as shown in Fig. 2. The outer perimeters ofthe six square apertures are drawn to scale as dashed lines onthe contour plot. The elongated extent of each single-elementSAR pattern along its feedline axis [from Fig. 6(a)] allows theuse of larger interelement separation in thedirection than inthe horizontal direction, fortuitously providing more spacefor running multiple microstrip feedlines between elementsof large array structures. For this interelement spacing of

7.5 mm and 1.5 cm, the 50% SAR contourline is contiguous and demonstrates excellent coverage of thesurface area under the entire array perimeter. Fig. 10 quantifiesthe percentage of array surface area circumscribed by all sixapertures attaining each specific normalized SAR value in theplane 1-cm-deep in muscle. Arrays with three different verticalelement spacings (7.5-mm, 1.25-cm, and 1.5-cm) are comparedusing 5-mm-thick silicone oil coupling bolus and two differentvertical spacings of elements (7.5-mm and 1.5-cm) with a5-mm-thick water bolus layer. Using the silicone oil bolus,50% or more of the maximum SAR is obtained over 98% ofthe array surface in this 1-cm-deep plane for all three verticalseparations tested. For the more commonly used water boluslayers, the SAR coverage is not quite as complete, especiallyfor elements spaced 1.5 cm apart which demonstrated only80% coverage 50% SAR . The same analysis performedin the horizontal plane 5-mm deep in muscle produces verysimilar results with over 90% of the area under the apertureabove 50% SAR with silicone oil coupled arrays at allvertical separations and 70% of the area covered in the worstcase of water bolus coupled 1.5-cm-spaced apertures.

Fig. 8. Comparison of ideal feeding of one gap with true single microstripfeedline excitation. Normalized SAR distributions in a horizontal plane 1-cmdeep in the muscle below a 5-mm-thick water bolus. Solid contour lines referto the ideal feed configuration while the dotted lines refer to the microstripfeedline.

IV. DISCUSSION

A DCC microwave antenna has been developed for use as abuilding block element of large multielement array applicatorsfor superficial hyperthermia of large contoured tissue regions.In previous work, square DCC radiators were constructed inseveral sizes from 3.5 to 6 cm across, and detailed-field mea-surements were made in muscle phantom to demonstrate thedesirable annular shell of power deposition emanating from theslot around the aperture periphery [10], [11]. Previous effortshave also demonstrated the accuracy of the FDTD method forsimulating power deposition from DCC aperture antennas bydirectly comparing theoretical FDTD simulations with-fieldmeasurements for single 3.5-cm-square apertures [11] and forsmall, three-, four-, and six-element, array configurations [10],[12]. The present effort extends the theoretical FDTD model-ing work to characterize several important features of the DCCantenna which will affect the practical design, interfacing, andclinical performance of large area conformal array applicatorsto be constructed from multiple closely spaced apertures.In order to provide applicators with finer resolution controlover the heating pattern to accommodate heterogeneous tissueproperties, this study investigates smaller 2-cm-square DCCradiators which will be driven independently with noncoherent915-MHz power sources. To eliminate awkward coaxial cablefeedline connections to each aperture of large arrays, weinvestigate several different microstrip feedline configurationswhich preserve the thin and lightweight structure of theconformal PCB applicator. And, finally, to study the effectof various aperture design and coupling layer changes on thenear-field radiation patterns, we present a relatively rare viewof the SAR patterns in a vertical plane cross section throughthe applicator, bolus, and tissue load.

Results of the current FDTD simulations demonstrate thatwhen a single microstrip feedline is used to bring microwavepower to the DCC aperture across only one side gap instead


(a)

(b)

Fig. 9. (a) Normalized SAR in a horizontal plane 1-cm deep in muscle for a six element DCC aperture array with each aperture fed by an ideal source acrossthe middle of its�y direction side gap and coupled to muscle with a 5-mm-thick silicone oil bolus. Spacing between apertures is 7.5 mm in thex directionand 1.5 cm in they direction. Accurate projections of aperture positions are indicated with dashed lines on the contour plot. (b) Comparison of 25%, 50%,and 75% SARmax contours for 5-mm-deep (solid) and 1-cm-deep (dashed) planes in muscle under 5-mm-thick silicone oil bolus.

of by the more difficult to realize equal amplitude and phasefour corner feed configuration, the radiation pattern in muscleis skewed significantly from the symmetric SAR of Fig. 4 tothat shown in Fig. 5. Fortunately, this asymmetry of SAR canbe minimized by adding a thin layer of dielectric betweenthe applicator and lossy tissue load, as shown in Fig. 6. Forthis 2-cm-square aperture, a 2.5–5-mm-thick layer of readilyavailable silicone oil fluid provided a rebalancingof and component -fields which recentered the SARunder the aperture. This same 5-mm bolus layer was then usedwith an array of six 2-cm-square DCC apertures spaced 7.5mm apart horizontally and 1.5 cm apart vertically to obtainthe very desirable SAR pattern shown in Fig. 9 and analyzedquantitatively in Fig. 10. For the intended application of large

area superficial hyperthermia treatments, this basic radiationpattern covering the entire area under the six-element arrayperimeter with SAR between 50% and 100% of maximumwill be expanded further by placing additional DCC radiatorstogether to cover the required area.

Because the practical application of this superficial heatingdevice requires the use of a temperature regulated bolus tocontrol skin temperature and allow slightly deeper penetrationof the heating field, we investigated the effect of varying bolusthickness from a very thin 2.5 mm to a bulky 2 cm. With nobolus, the field was highly concentrated in a few millimetersof tissue just under the gap around the aperture periphery,and especially under the excited ( axis) gap. As seen inFig. 7 however, SAR extended deeper and more centrally


(a)

(b)

Fig. 10. SAR/Area histograms showing the percent area under the sixaperture array perimeter that is covered by various SAR levels in the horizontalplane: (a) 1-cm deep and (b) 5-mm deep in muscle. Vertical spacings of7.5 and 15 mm between adjacent apertures were considered for the cases of5-mm-thick water bolus and 5-mm-thick silicone oil bolus layers.

under the aperture with progressively thicker bolus layers dueto divergence of the near field. While similar smoothing andcentering of the SAR pattern was reported in previous studies[16] using deionized water bolus , we found thatultra-low viscosity 50-cS silicone oil provided a coupling fluidwith properties quite similar to water except for the lowerdielectric constant which was better matched tothat of the PCB substrate. Due to the corresponding increasein effective wavelength of radiation from about 3.6 in waterto 20 cm in silicone oil, the latter low dielectric constantmaterial provided additional expansion, smoothing and center-ing of the radiated field under the aperture. Because thermalconduction redistribution of heat within a perfused tumor willalso tend to smooth the temperature distribution which resultsfrom nonuniform power deposition, the ideal SAR pattern forDCC apertures is considered to be one with some remainingperipheral enhancement of the SAR superficially expanding tofull coverage under the aperture footprint at the intended depth

of penetration (1 cm in soft tissue). Thus the SAR patternsobtained with conveniently thin 2.5–5-mm silicone oil bolus[Fig. 7(b) and (c)] appear more suitable than the centrallypeaked power deposition patterns characteristic of thickerlayers [Fig. 7(e) and 7(f)] and previously used waveguide-type microwave applicators [17]. Future efforts will be directedat coupling this electromagnetic analysis to a separate FiniteDifference thermal modeling program to provide additionalconfirmation of the optimum bolus thickness and temperaturefor most uniform tissue heating under arrays of DCC apertures.

V. CONCLUSION

The FDTD numerical method has been used to investigateradiation patterns from 915-MHz DCC applicators in both sin-gle aperture and six-element array configurations. The methodhas proved useful for theoretical characterization of applicatorperformance and optimization of applicator design parametersfor various electromagnetic feed excitation and tissue loadingconditions. Results demonstrate the feasibility of replacingcoaxial cable feedlines to each radiating aperture with lessbulky microstrip feedlines printed on the back surface of thindouble-sided PCB material and extending across just one sidegap of each aperture. Results also demonstrate that while acoupling bolus layer is required to obtain an SAR pattern thatis appropriately localized to the area under the aperture, theideal bolus thickness to maintain the desirable slight peripheralenhancement of SAR is between 2.5–5 mm which preservesthe thin and lightweight nature of the applicator. Analysis ofsix-element array configurations shows good uniformity ofSAR within the array boundaries and extension of the 50%SAR contour to the periphery of the array footprint inboth the 5-mm- and 1-cm-deep planes for several practicalinterelement spacings using 2-cm-square apertures and 5-mm-thick silicone oil coupling bolus. The simulations clearlydemonstrate that uniform heating of large superficial tissueregions can be successfully realized by using lightweight andflexible multi-aperture DCC array applicators made from thinPCB material coupled to tissue with a 2.5–5-mm-thick bolus.

ACKNOWLEDGMENT

The authors would like to acknowledge the invaluablecontribution of the University of Florence in the developmentand training of FDTD analysis techniques, and previous effortsof V. Manfrini.

REFERENCES

[1] D. S. Kapp, “Efficacy of adjuvant hyperthermia in the treatment ofsuperficial recurrent breast cancer: Confirmation and future directions,”Int. J. Radiat. Oncol. Biol. Phys., vol. 35, no. 5, pp. 1117–1121, 1996.

[2] W. C. Dewey, “Arrhenius relationships from the molecule and cell tothe clinic,” Int. J. Hyp., vol. 10, no. 4, pp. 457–483, 1994.

[3] P. K. Sneed and T. L. Phillips, “Combining hyperthermia and radiation:How beneficial?,” Oncol., vol. 5, pp. 99–108, 1991.

[4] J. R. Oleson, T. V. Samulski, K. A. Leopold, S. T. Clegg, M. W.Dewhirst, R. K. Dodge, and S. L. George, “Sensitivity of hyperthermiatrial outcomes to temperature and time: Implications for thermal goalsof treatment,”Int. J. Radiat. Oncol. Biol. Phys., vol. 25, pp. 289–297,1993.

[5] C. C. Vernon, J. W. Hand, S. B. Field, D. Machin, J. B. P. van Dijk,D. Gonzalez Gonzalez, J. J. Liu, P. Goodman, and M. Sherar, “Ra-


diotherapy with or without hyperthermia in the treatment of superficiallocalized breast cancer: Results from five randomized controlled trials,”Int. J. Radiat. Oncol. Biol. Phys., vol. 35, no. 4, pp. 731–744, 1996.

[6] J. Overgaard, D. G. Gonzales, M. C. C. H. Hulshof, G. Arcangeli,O. Dahl, O. Mella, and S. M. Bentzen, “Hyperthermia as an adjuvantto radiation therapy of recurrent or metastatic malignant melanoma. Amulticentre randomized trial by the european society for hyperthermiaoncology,” Int. J. Hyp., vol. 12, no. 1, pp. 3–20, 1996.

[7] C. A. Perez, B. Gillespie, T. Pajak, N. B. Hornback, B. Emami, andP. Rubin, “Quality assurance problems in clinical hyperthermia andtheir impact on therapeutic outcome: A report by the radiation therapyoncology group,”Int. J. Radiat. Oncol. Biol. Phys., vol. 16, pp. 551–558,1989.

[8] C. C. Johnson and A. W. Guy, “Nonionizing electromagnetic waveeffects in biological materials and systems,” inProc. IEEE, vol. 60, pp.692–718, 1972.

[9] Thermoradiotherapy and Thermochemotherapy, Biology, Physiology,and Physics, vol. 1, M. H. Seegenschmiedt, P. Fessenden, and C. C.Vernon Eds. Berlin, Germany: Springer-Verlag, ch. 10, 1995.

[10] P. R. Stauffer, C. J. Diederich, and D. Bozzo, “Conformal arraymicrowave applicator for superficial hyperthermia of large contouredsurfaces,”1994 IEEE MTT-S Int. Microwave Symp. Dig., 1994, vol. 1,pp. 531–534.

[11] P. R. Stauffer, M. Leoncini, V. Manfrini, G. B. Gentili, C. J. Diederich,and D. Bozzo, “Dual concentric conductor radiator for microwavehyperthermia with improved field uniformity to periphery of aperture,”IEICE Trans. Communicat., vol. E78-B, no. 6, pp. 826–835, 1995.

[12] V. Manfrini, P. R. Stauffer, M. Leoncini, and C. J. Diederich, “Dualconcentric conductor arrays for microwave hyperthermia: Theoreticalstudy of design parameters,” inProc. Wescon/95, San Francisco, 1995,pp. 662–667.

[13] K. S. Yee, “Numerical solution of initial boundary value problems in-volving Maxwell’s equations in isotropic media,”IEEE Trans. AntennasPropagat, vol. AP-14, pp. 302–307, 1966.

[14] A. Taflove and K. Umashankar, “The finite-difference time-domain (FD-TD) method for electromagnetic scattering and interaction problems,”J. Electr. Waves and Appl., pp. 243–69, 1987.

[15] A. Taflove, “Application of the finite-difference time-domain methodto sinusoidal steady-state electromagnetic penetration problems,”IEEETrans. Electromag. Compat., vol. EMC-22, pp. 191–202, 1980.

[16] F. Rossetto, P. R. Stauffer, C. J. Diederich, D. Deardorff, G. B.Gentili, and M. Leoncini, “FDTD characterization of radiation patternsfrom flexible microstrip applicators,” inProc. IEEE Antennas andPropagation Societry Symp. Dig., Montreal Canada, 1997, vol. 3, pp.1562–65.

[17] W. L. Straube, R. J. Myerson, B. Emami, and L. B. Leybovich, “SARpatterns of external 915 MHz microwave applicators,”Int. J. Hyp., vol.6, no. 3, pp. 665–670, 1990.

Paul R. Stauffer received the B.A. degree inphysics at the College of Wooster, Wooster, OH, in1975 and the M.S. degree in electrical engineeringat the University of Arizona, Tucson, AZ, in 1979.

He is currently Associate Adjunct Professor ofRadiation Oncology at the University of California,San Francisco and Director of Physics andEngineering for Hyperthermia. His current researchinterests include the engineering development andtesting of RF, microwave and ultrasound applicatorsfor uniform heat treatment of superficial and deep

lying tissues. He has published over 80 papers, reports, and book chaptersin the field of hyperthermia, and is an Associate Editor for theInternationalJournal of Hyperthermiaand theInternational Journal of Radiation OncologyBiology and Physics.

Dr. Stauffer is board certified in clinical engineering (1983) and medicalphysics (1991).

Francesca Rossettowas born in Milano, Italy,on May 16, 1970. She received the Dr. Degreein electronics engineering from the University ofFlorence, Florence, Italy, in 1996.

She is now working as a Research Fellow in theRadiation Oncology Department at the University ofCalifornia, San Francisco, on theoretical modelingof microwave hyperthermia applicators using theFDTD method.

Marco Leoncini was born in Pontremoli (MS) Italyon January 16, 1960. He received the Dr. degree inphysics from the University of Pisa, Pisa, Italy, in1984.

From April 1985 to October 1986, he was in theItalian Navy as a Reserve Officer in ElectromagneticCompatibility. From October to December 1990,he was a Visiting Researcher in the Department ofRadiation Oncology, University of Arizona, Tucson.Since January 1991, he has been a Technical Man-ager in the Department of Electronics Engineering,

University of Florence, Florence, Italy. His current research interests includetheoretical analysis of wave propagation in inhomogeneous and dispersivematerials and numerical solution of Maxwell’s equations using the finite-difference time-domain method.

Dr. Leoncini was awarded a grant in March 1987 from SMA Inc. in theDepartment of Electronics Engineering at the University of Florence, for theanalysis and design of microwave applicators for intracavitary hyperthermia.

Guido Biffi Gentilli was born in Lucca, Italy, onAugust 9, 1943. He received the Dr. degree inelectronics engineering from the University of Pisa,Pisa, Italy, in 1970.

From 1970–1976, he was with the Department ofElectrical Engineering, University of Pisa. In 1977,he joined the Department of Electronics Engineeringat the University of Florence, Florence, Italy, wherehe teaches courses on radar theory and techniques.He is currently an Associate Professor of MicrowaveCircuit Design. His research interests are in the area

of microwave remote sensing, CAD modeling of microwave devices, biolog-ical applications of EM waves, and numerical methods in electromagnetics.

Dr. Gentili is a member of the Italian Electrotechnical and ElectronicAssociation (AEI).

Documents

Radiation patterns of dual concentric conductor microstrip antennas for superficial hyperthermia