A Coaxial Antenna With Miniaturized Choke for Minimally Invasive Interstitial Heating

Transcription

1 82 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 1, JANUARY 2003 A Coaxial Antenna With Miniaturized Choke for Minimally Invasive Interstitial Heating Iginio Longo, Guido Biffi Gentili, Matteo Cerretelli, and Nevio Tosoratti* Abstract We present a new coaxial antenna for microwave interstitial coagulative therapy, working at 2450 MHz and endowed with a miniaturized sleeve choke in order to reduce back heating effects and make the system response less dependent on the antenna insertion depth into the tissue; the way the choke is implemented makes the overall transversal size minimum and allows small adjustments of the choke section length even during operation. We describe the main technical features of the antenna and show experimental results clearly proving the choke effectiveness. Numerical simulations well agree with experimental data, confirming the suitability of the proposed device for minimally invasive medical applications. Index Terms Interstitial coaxial applicator, microwave coagulative therapy, miniaturized choke. I. INTRODUCTION THE demand for minimally invasive interstitial heating techniques employing electromagnetic sources has been constantly increasing in recent years, particularly in the oncological field. In a variety of medical applications, such as the thermal ablation of solid tumours, the use of microwave probes is under many respects preferable to that of radio-frequency (RF) current loops or laser beams, due to a greater heating effectiveness (in terms of heated volume per unit time and per unit applied power) and/or homogeneity. Indeed, microwaves and RF currents show about the same heating capacity (with a slight prevalence of the former, as shown by Lin [1]), much greater than that of laser beams, characterized by a very small penetration depth (8 mm at most in perfused tissues, using a multiscattered Neodymium:Yttrium Aluminum Garnet source, with a wavelength of 1064 nm [2]). Besides, microwaves are less sensitive than RF currents to local variations in the conductivity of the tissue, resulting in a more uniform heating. In addition, microwave probes generally exhibit lower power densities than their RF and laser counterparts, reducing the risk of tissue charring close to the emitter. Microwave interstitial applicators are, thus, very interesting tools for minimally invasive surgery. However, conventional coaxial antennas of the asymmetric dipole type (often referred to as monopole antennas [3]) need Manuscript received February 28, 2002; revised August 25, This work was supported by Consiglio Nazionale delle Ricerche (CNR). Asterisk indicates corresponding author. I. Longo is with IPCF, CNR Area della ricerca di Pisa, Pisa, Italy. G. B. Gentili and M. Cerretelli are with Dipartimento di Elettronica e Telecomunicazioni, Università degli studi di Firenze, Firenze, Italy. *N. Tosoratti is with IPCF, CNR Area della ricerca di Pisa, Via G. Moruzzi 1, Pisa, Italy ( nevio@ifam.pi.cnr.it). Digital Object Identifier /TBME proper design modifications in order to minimize undesired effects due to exceeding or uncontrolled power reflection at their open end. The amount and spatial distribution of the reflected power strongly affect the form and size of the heating pattern: indeed, the currents flowing from the antenna active section back to the microwave source along the outer coaxial conductor give the heating pattern a typical tear drop shape (as confirmed by lab trials both on autoptic liver and in vivo, supported by computerized tomographic imaging techniques [4]), which undergoes considerable changes as the antenna insertion depth into the tissue is varied [3]. Though backward heating is sometimes useful, e.g., to stop residual bleeding and to kill tumoural cells all along the antenna, it is generally avoided both in thermoablative techniques, which are minimally invasive and most effective when the heating pattern covers the lesion area only, and in percutaneous microwave coagulative therapies, for which the production of skin burns at the antenna insertion point is utterly undesired. A practical solution to minimize back heating effects and to make the performances of the antenna less dependent on its depth of penetration into the tissue is the introduction of a metallic choke surrounding the coaxial cable, capable of trapping the reflected wave [5] [8]. The insertion of the choke inevitably causes the overall diameter of the applicator to increase, which makes the interstitial heating procedure more invasive: therefore, for a given coaxial antenna (whose size is fixed depending on the power it must handle), the main challenge in choke design is keeping the transversal size at a minimum. In this paper, we present a new, minimally invasive coaxial antenna featuring a miniaturized choke of simple design; its most original feature is in that the biopsy needle used to introduce the antenna into the lesion to be ablated serves also as lateral metallic wall of the choke section: this solution attains minimum overall transversal size and allows small real time adjustments of the choke section length to compensate for little impedance mismatches which may arise during operation. We describe the radio-electrical characteristics of the proposed device and show several experimental results obtained on lab phantoms, which, compared to the performance of an unchoked replica of the same applicator, clearly prove the choke effectiveness. Experimental data are in pretty good agreement with the results of a numerical analysis based on simplified electromagnetic models of our choked and unchoked applicators, making use of the Method of Moments (MoM) [9]. II. DESIGN AND REALIZATION OF THE CHOKED APPLICATOR Fig. 1(a) provides the longitudinal cross-sectional view of a conventional asymmetrical dipole coaxial antenna endowed /03$ IEEE

2 LONGO et al.: COAXIAL ANTENNA WITH MINIATURIZED CHOKE FOR MINIMALLY INVASIVE INTERSTITIAL HEATING 83 Fig. 1. (a) Cross section of a conventional choked coaxial antenna of the asymmetric dipole type. Legend: EC=CC = external/central coaxial conductor (copper), I = insulator (P.T.F.E.), PC = plastic catheter (e.g., silicone), CS = choke section, F = antenna feed, T = antenna tip, DT = dielectric tip; (b) cross section of a prototype of our new choked applicator; all sizes are in mm. Legend: BN = biopsy needle, C = copper collar, S = solder, PT = plastic tubing (P.T.F.E.), L = length of CS, L = distance between CS input and T, L = distance between CS input and F, L = length of DT; (c) Equivalent circuit of our prototype for MoM analysis. Legend: AA = reference plane at CS input, BB = reference plane at feeding point, LS = localized feeding source at the BB section, G = feeding generator, SS = subsection of the antenna having an impedance equal to that of the choke at AA, and BC = bulk conductor. Figures are not to scale. with a sleeve choke [5], to be compared with the design of our new applicator in Fig. 1(b). In both schematic representations, EC/CC denotes the external/internal coaxial conductor, designates the coaxial insulator, and CS indicates the choke section, that is a quarter wavelength short-circuited coaxial line behaving as an open circuit at its input. In Fig. 1(a), CS is obtained from a monolithic metal cap, the entire applicator being encapsulated into a plastic catheter,(pc), which guides the antenna as deep as required into the tissue. In Fig. 1(b), PC is substituted by a metal biopsy needle (BN) which is short circuited to EC through a metal collar,, soldered onto the cable: is the end wall of the choke section, whose lateral shielding is provided by BN itself. Note that incorporating the choke metallic walls into the guiding structure (BN in our case) makes the transversal size of the device minimum: with respect to Fig. 1(a) a reduction of the overall diameter equal to twice the thickness of the metal cap is obtained. With our design variant the limit to miniaturization is, thus, fixed only by power handling requirements, which impose a lower bound to the coaxial cable diameter. The geometrical details reported in Fig. 1(b) refer to a prototype of our choked applicator working at 2.45 GHz. The antenna was made out of a semirigid coaxial cable (Huber Suhner AG, Herisau, CH, type EZ 47 M 17), 25 cm long, 1.19 mm wide 0.047, with copper conductors and filled with solid P.T.F.E. (relative permittivity: ). The cable had the following nominal electrical characteristics: 50 characteristic impedance; 35 W of average handling continuous wave (CW) power at 40 C, with VSWR 1.0; maximum attenuation of 170 db/100 m at 20 C. The coaxial antenna was endowed with an SMA male connector [not shown in Fig. 1(b)] soldered at the proximal end, while the opposite (distal) end was left unshielded by peeling off the outer conductor, so as to form a dielectric tip (DT) 16 mm long. Throughout this paper, we shall refer to the distal end of the outer conductor as feed (point),,or junction, while the distal end of the central conductor will be indicated as tip,. The miniaturized sleeve choke [10] was made out of a copper collar with 1.2-mm inner diameter, 1.78-mm outer diameter and a few millimeters long, soldered onto EC. A thin-walled plastic tube (PT) (in solid P.T.F.E.), extending from to was inserted tight around EC and the whole assembly was eventually fit snugly into a 14-gauge (14-G) metal biopsy needle (inner diameter: 1.78 mm, outer diameter: 2.05 mm), BN. A good sliding electrical contact was obtained between the copper collar and the inner surface of the needle, with PT acting as a water tight, low-friction, centering cushion. Note that the length of the choke section, whose nominal value is mm, may be slightly adjusted even during operation by letting BN slide onto : this additional degree of freedom is another original feature of our applicator and proves useful to compensate for small deviations from optimal working conditions, e.g., due to little changes in the permittivity of the heated tissue or of the insulating tube PT with increasing temperature. It is worth noting the ease of construction of the proposed device and the wide availability and low cost of all its components. By comparison, consider the choked applicator presented in [7], made out of a triaxial cable UT (Micro-coax Components, Collegeville, PA), far more expensive than our ordinary 50 coaxial cable (by at least one order of magnitude): it has a somewhat larger overall diameter (2.22 mm versus 2.05 mm) and, if it were made to work at 2.45 GHz like ours, the average CW power supported would be less than half (16 W versus 35 W) [11]. All experiments described successively were performed using the choked prototype of Fig. 1(b) and, for direct comparison, an unchoked applicator of the same type and structure, that is, made out of the same coaxial cable, having the same dielectric tip length and the same overall length, but devoid of the metal collar. The electromagnetic models used to simulate the response of both applicators are strictly based on the geometry and material information contained in Fig. 1(b). III. ELECTROMAGNETIC MODELLING AND NUMERICAL PROCEDURE For the numerical analysis of the choked applicator of Fig. 1(b) and of its unchoked replica, a software based on the MoM was used [12]. The MoM integral equation approach [9] proves very advantageous with respect to finite-element (FE) or finite-difference time-domain (FDTD) methods when dealing with rather simple and slender radiating elements, such as interstitial applicators [13], [14] operating in a medium whose Green function is known. The reason is that MoM does not

3 84 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 1, JANUARY 2003 require discretization of the analysis domain and reduces the unknowns to the sole currents flowing on the metallic and/or dielectric surfaces of the radiating structure. Actually, the use of Berenger s perfectly matched layer (PML) [15] instead of the traditional absorbing boundary conditions (ABC) [16] generally yields a considerable reduction of the FDTD analysis domain; however, it does not help reducing the computational cost when the dimensions of the analysis domain are to be chosen independently of the PML criteria, as is the case for antenna near field calculations. In our specific case, we verified that for a sufficiently accurate FDTD near field analysis a mesh made up of more than cells is required, while for our MoM-based analysis it suffices to discretize the antenna with about 80 wire segments only; as a result, the MoM algorithm runs about 60 times faster than its FDTD counterpart. The simplified model used for the MoM analysis of our choked applicator is sketched in Fig. 1(c) and described hereafter. An equivalent localized feeding source (delta gap), LS, is collapsed at the junction or feed section, BB, followed by a thin metal wire with dielectric coating, representing the dielectric tip (DT) section; the use of a properly defined equivalent surface impedance for the radiating wire DT accounts for the presence of the insulating cladding [9]. The entire choke section (CS) is substituted with a proper impedance subsection, SS, collapsing at CS input, that is plane AA. The effective impedance of SS is made up of two contributions: the first,, is the impedance seen from AA looking toward the short circuited choke section and is readily available from standard transmission line formulas or using the Smith chart ( when ); the second,, is the impedance seen from AA looking toward the applicator tip. The determination of is not as straightforward as that of, since fringing field effects at AA, featuring a sharp structural discontinuity, must be accounted for: a MoM technique provided a fairly accurate evaluation of, namely by calculating the input impedance of a piece of coaxial line equal in diameter to CS, with a protruding cladding section having the same overall length as the radiating structure embedded into the tissue [, using the symbols defined in Fig. 1(b)]. Finally, the parallel of and yields the total impedance of subsection SS. The rest of the structure (everything to the left of AA) is simply modeled as a metallic wire, BC. It is worth noting that if the nonmetallic parts of the analysis domain were discretized using dielectric cuboids, the numerical efficiency of the MoM algorithm would have been strongly penalized [17]. The model used to analyze the unchoked applicator differs from that in Fig. 1(c) only for the absence of the impedance subsection SS: indeed, in the latter case no discontinuities appear prior to the antenna feed section BB and the whole coaxial line to the left of BB may be treated as a mere metallic wire, BC, leaving the rest of the model unchanged. Our strongly simplified electromagnetic models produce approximated results, whose accuracy, however, may be evaluated through a proper numerical sensitivity analysis. For example, consider the effects of the rather arbitrary sizing of subsection SS, which was given a length to diameter ratio of about 2 in compliance with specific software requirements relative to the meshing of the structure: we explicitly verified that varying the Fig. 2. A schematic representation of the experimental setup: MS = microwave source (Opthos Inc. Mod. MPG4), MA = microwave applicator, EMS = electromechanical switch, FOT = fiberoptic thermometer, FS = fluoroptic sensor, OF = optical fiber, VNA = vector network analyzer (Hewlett Packard, Mod B), EW = egg white, CC = coaxial cable, DL = dummy load, GB = glass beaker, and MJ = mini-jack. For simplicity, the power meter (Bontoon, Mod. TP 01) and the directional coupler (Narda, Mod. 3022) used to measure the power actually delivered to the antenna are not shown. length of SS does not appreciably affect the calculated antenna field distribution (provided the surrounding tissue is homogeneous) and only slightly affects the calculated input reflection coefficient ( changes only by 4% upon doubling the subsection length). Much the same may be said relative to the sizing of the feeding gap at the pertinent BB section in Fig. 1(c). In all, none of the model parameters which had to be fixed empirically proved to be critical. The described models allow the calculation of the antenna input reflection coefficient, of the near field distribution, of the specific absorption rate (SAR) pattern and of several other relevant quantities. IV. EXPERIMENTAL APPARATUS Fig. 2 shows a sketch of the apparatus used for the experimental characterization of our microwave applicator (MA). The microwave source (MS) is a magnetron oscillator working at 2450 MHz, with a maximum CW output power of 100 W. The input power to MA, delivered only during preselected time intervals through an electromechanical switch (EMS), is measured by a power sensor mounted on a bidirectional coupler (both not shown in Fig. 2). The temperature is measured by a fiberoptic thermometer (FOT) coupled to a fluoroptic sensor (FS) of about 1.1 mm in diameter, with a characteristic reading time of about 0.5 s (in water) and an accuracy of 1 C. A vector network analyzer (VNA), performs scattering parameter measurements as a function of frequency and of the antenna insertion depth into static aqueous phantoms: egg white (EW) was preferred to water as it coagulates, allowing easy visualization of the heating pattern, and is less subject to fast convective heat

4 LONGO et al.: COAXIAL ANTENNA WITH MINIATURIZED CHOKE FOR MINIMALLY INVASIVE INTERSTITIAL HEATING 85 Fig. 3. Numerically computed power density distribution (db 1 W/m ) produced by the choked (a) and unchoked (b) applicator radiating in a homogeneous medium with relative permittivity equal to that of EW (" =690 j 14.5 at 2.45 GHz). A more pronounced bending of the iso-power lines (corresponding to iso-sar lines) clearly shows close to the feed section when the choke is present. The core of the SAR pattern produced by the choked antenna extends for the most part between the feed and the tip of the antenna, indicating the effectiveness of the choke in blocking backward RF currents. transfer. Successive pictures of heating patterns in EWs were taken through a glass cell (not shown) endowed with a 6 cm 6 cm glass flat window. V. RESULTS A. Numerical Results Fig. 3(a) and (b) directly compares the calculated spatial distribution of the power density produced, respectively, by the choked applicator depicted in Fig. 1(b) and by its unchoked replica; calculations were performed assuming (here and everywhere else in this section) the medium surrounding the antenna to be homogeneous, with relative electric permittivity equal to that of EW (namely, 14.5 at 2.45 GHz, ambient temperature [18]). The equivalent impedance of subsection SS substituting the distributed choke section in the simplified model of Fig. 1(c) was numerically evaluated through the procedure outlined in Section III. The frequency dependence of the real and imaginary parts of the contributions and to are shown in Fig. 4(a): is the input impedance of a quarter wave transformer, resonating at 2.45 GHz; instead, accounts for fringing field effects at the geometrical discontinuity in AA an has a negative imaginary part. The parallel of and gives the total impedance, exhibiting a frequency dependence similar to that of a low quality factor resonant circuit, as shown in Fig. 4(b). Using the simplified model of Fig. 1(c) into which the calculated value of was inserted, we eventually evaluated the input reflection coefficient of our choked and unchoked applicators: the results are reported in Fig. 4(c). For both antennas turns out to be less than 10 db in the 1.7- to 3-GHz range: such a good and broadband impedance matching surpasses the most compelling requirements for interstitial applicators; the calculated reflection level is so low that the use of a circulator to protect the microwave source from reflected waves is virtually unnecessary. Besides, the choked antenna is expected to provide Fig. 4. Numerical results at 16 different frequencies (uniformly distributed in the 1.7- to 3-GHz range) for applicators immersed in a homogeneous medium equivalent to EW (" =690j 14.5 at 2.45 GHz and at ambient temperature): (a) frequency dependence of the real and imaginary parts of the impedance Z, seen from the outer coaxial conductor end [plane AA in Fig. 1(c)] looking toward the choke end-wall, and of the impedance Z, seen from AA looking toward the tip of the choked applicator. NB: note the different scales; (b) frequency dependence of the real and imaginary parts of the total impedance Z at plane AA, given by the parallel of Z and Z ; (c) computed reflectance (js j) of the choked and unchoked applicators; a third curve is also shown relative to an applicator with choke section shortened by 2 mm (a 10% reduction with respect to the nominal length at 2.45 GHz). a further impedance match improvement in the 2- to 2.7-GHz range: at our operating frequency, 2.45-GHz, reflections are expected to decrease by about 3 db. It is also worth noting that the actual choke section length seems not to be too critical to the overall choke performance: as shown in Fig. 4(c), even upon reducing by 2 mm, which is about 10% of the nominal choke section length at our operating frequency, the choked applicator is still almost optimally matched. Broadband operability and insensibility to relatively large mechanical tolerances make the device performances pretty stable, as well as easily predictable and reproducible. B. Experimental Results and Comparison With Predictions SAR measurements were performed by applying microwave pulses of high power and short duration to the choked and un-

5 86 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 1, JANUARY 2003 Fig. 5. Comparison between measured and numerically calculated SAR for both choked and unchoked applicators: each set of data is normalized to its peak value. Experimental data were obtained by sending pulses of up to 50 W of amplitude and 2 s of duration to applicators immersed in EW and by measuring the corresponding temperature increase 2 mm apart from the antenna axis. Numerical data were extracted from Fig. 3(a) and (b) and, thus, represent normalized power densities obtained from a MoM near field analysis based on the simplified electromagnetic model of Fig. 1(c). choked applicators, immersed in EW, and by measuring the corresponding local temperature increase [19]; the latter quantity coincides with SAR apart from a numerical factor. Temperature variations were recorded 2 s after the end of the microwave pulse: such a delay proved short enough to ignore heat waves coming from nearby regions in the liquid phantom at different temperatures, but still large with respect to the characteristic response time of the temperature sensor. Pulses up to 50 W of amplitude and 2 s of duration at a frequency of 2450 MHz were applied. Fig. 5 reports the temperature increase produced by the choked and unchoked antennas, measured at 2 mm from the radiator axis, as a function of the axial coordinate ( 0 at the feed position and increasing toward the tip), normalized to the respective peak values; for both applicators the maximum detected was 8 C, located at about 5 mm from the feed, in direction of the tip. In the same figure, we also report the power densities computed for both applicators, normalized to the respective absolute maxima: these data were extracted from Fig. 3. The normalized experimental and numerical data correspond to SAR profiles(in arbitrary units) and show very good agreement for all values of, save for a small region close to the feed where the choke effect appears overestimated with respect to actual performance: indeed, many spurious effectsnot included in our simplifiedelectromagnetic model (e.g., ohmic losses on the metallic walls and in the filling dielectric of the choke section, parasitic contact impedances relative to the soldered end wall, etc.) may well account for these discrepancies. The numerical curves as well as measured data relative to the choked and unchoked applicators are very close to each other in the distal zone and near the feed, whereas past the feed, for mm, an appreciable reduction of SAR values induced by the choke is observed, once again confirming its effectiveness. Fig. 6 collects successive pictures of coagulation patterns produced in EW (initially at 22 C) by applying a constantmicrowave power of 25 W, using both the unchoked antenna [Fig. 6(a)] and our new choked applicator [Fig. 6(b)]; arrows and indicate, respectively, the position of the antenna feed and of the choke end wall (copper collar), while FS designates the fluoroptic temperature sensor. Direct Fig. 6. Successive pictures of the coagulum produced in EW by the unchoked antenna (a) and by the antenna with miniaturized choke (b): in both cases the applied power was P = 25 W and the initial temperature of EW was T =22 C. F designates the feed point, C indicates the choke end wall (copper collar) and FS represents the fluoroptic temperature sensor. Pictures taken after 15 and 45 s provide direct comparison of the heating patterns relative to the choked and unchoked applicators. Pictures taken after 60 s for the unchoked applicator and after 120 s for the choked one catch almost definitive qualitative features of the coagulation process: indeed, protracted heating causes the coagulated volume to expand at the same rate in all directions, so that the relative dimensions of the characteristic ellipsoidal coagulum are about constant. comparison of pictures taken after 15 and 45 s unequivocally show the choke effectiveness: the coagulum produced by the unchoked applicator has the characteristic tear drop shape and encloses the feed, while that produced by the choked applicator is contained for the most part in the half space below the feed. In both cases, when heating is protracted the coagulum forms a well defined and reproducible ellipsoidal pattern, similar in volume for the choked and unchoked applicators, though less stretched when the choked antenna is in use. It was observed that heating treatments as short as 1 2 min yet retain all relevant qualitative features of the coagulation process: by that time, the coagulated ellipsoid is fully formed and, though enlarging upon further heating, its semi-axes appear to maintain an almost constant ratio. The temperature ranges from 65 C 1 C inside the coagulated volume, at 1 mm from its external margin, to 62 C 1 C just outside the coagulum,

6 LONGO et al.: COAXIAL ANTENNA WITH MINIATURIZED CHOKE FOR MINIMALLY INVASIVE INTERSTITIAL HEATING 87 change as temperature increases: slight adjustments of may then be exploited to compensate for small mismatches arisen during operation and avoid, to some extent, deterioration of the choke performance. Upon displacing the needle with respect to the antenna by about 1 mm, we tracked a 5% reduction in the reflected power; the best result obtained was a 8% reduction of with a 2-mm displacement of the needle. Such experimental findings go well together with the numerical results displayed in Fig. 4(c), showing that when our applicator is operated at 2450 MHz in a lossy medium (such as EW) a 10% deviation of the choke length from its nominal value,, still complies with a good impedance match. Fig. 7. Power reflected by the choked (P, squares) and by the unchoked (P, crosses) antenna as a function of the depth of insertion (d) into the EW, with 30-W incident power. Note that the choked antenna is already matched for d =16mm (that is, with the antenna feed at the liquid surface) and remains matched for deeper insertions, while the unchoked antenna is not matched (and exhibits a rapidly changing performance as the antenna is inserted deeper in the phantom) until d =25mm. In the inset, the ratio P =P is shown: note that its average value for deep immersions (d 25 mm), that is 03 db, is readily comparable to that calculated numerically [see Fig. 4(c)]. 1mm apart; the uncertainty on temperature values is mostly due to non homogeneities of the material, to the sensor size and to residual experimental errors. We also measured the power reflected by the choked antenna,, and by the unchoked antenna,, as a function of the depth of immersion into EW ( when the tip is just in touch with the liquid surface), with 30 W incident power. As shown in Fig. 7, in all operating conditions we obtained:. We observed for both applicators that, as is increased, the reflected power first decreases, then reaches a minimum and finally settles on an almost constant value. However, the choked antenna showed a considerably wider -independent zone of operation, starting already at mm (that is, with the antenna feed placed at the liquid surface): the latter feature of our device may be particularly attractive for the treatment of superficial lesions. The values of measured in full immersion conditions very well match (in average) the numerical prediction, namely 3 db [see, also, Fig. 3(c)]. In all, the experimental data confirm that our new applicator mainly radiates in forward direction: indeed, if the radiation pattern of the choked antenna were not a lobe extending forward for the most part, then it would be largely affected by reflections at the phantom/air boundary which, in turn, strongly vary with the insertion depth, resulting in sensibly -dependent device performances. The choked antenna was also used to coagulate large volumes of autoptic pig liver and a few interesting results were obtained. First of all, the plastic lining PT completely avoided adhesion of the applicator to the tissue, even upon protracted heating: extraction of the applicator at the end of the treatment was, thus, extremely facilitated. We also verified that fine dynamic adjustments of the length of the choke section, allowed by the sliding contact between the end wall of the choke and the guiding biopsy needle, produced small but appreciable improvements in impedance matching. Indeed, the permittivity of the insulator filling the choke section, as well as that of the heated tissue, inevitably VI. CONCLUSION We presented a new coaxial antenna for minimally invasive microwave interstitial coagulation therapy, working at 2.45 GHz and featuring a miniaturized choke integrated into a guiding biopsy needle. Our design solution attains minimum overall transversal size by letting the needle serve also as lateral metallic wall of the choke section; moreover, as the needle is free to slide with respect to the coaxial cable, real time adjustments of the choke section length are allowed: both these features are thoroughly original in the field of choked interstitial applicators design. We provided a numerical and experimental characterization of our choked antenna, verifying its effectiveness in comparison with an unchoked applicator of the same type: a good agreement was found between SAR and input reflection coefficient measurements and the corresponding numerical predictions. Our device operates in nearly perfectly matched conditions over a broad frequency band and its performances are not critically dependent on the choke section length and on the depth of insertion into the heated tissue within relatively large intervals of variability of the latter parameters. Easy realization and low cost make the proposed device an interesting candidate to the industrial production of interstitial applicators for the treatment of both deep and superficial lesions with minimally invasive surgery techniques. REFERENCES [1] J. C. Lin, Y. L. Wang, and R. J. Hariman, Comparison of power deposition patterns produced by microwave and radiofrequency cardiac ablation catheters, Electron. Lett., vol. 30, no. 12, pp , [2] J. T. De Sanctis, S. N. Goldberg, and P. R. Mueller, Percutaneous treatment of hepatic neoplasms: A review of current techniques, Cardiovasc. Intervent. Radiol., vol. 21, pp , [3] S. Labontè, A. Blais, S. R. Legault, H. O. Ali, and L. Roy, Monopole antennas for microwave catheter ablation, IEEE Trans. Microwave Theory Tech., vol. 44, pp , Oct [4] K. Mitsuzaki, Y. Yamashita, T. Nishiharu, S. Sumi, T. Matsukawa, M. Takahashi, T. Beppu, and M. Ogawa, CT appearance of hepatic tumors after microwave coagulation therapy, AJR, vol. 171, pp , [5] R. D. Nevels, G. D. Arndt, G. W. Raffoul, J. R. Carl, and A. Pacifico, Microwave catheter design, IEEE Trans. Biomed. Eng., vol. 45, pp , July [6] J. C. Lin, K. J. Beckman, R. J. Hariman, S. Bharati, M. Lev, and Y. L. Wang, Microwave ablation of the atrioventricular junction in openchest dogs, Bioelectromagnetics, vol. 16, pp , [7] W. Hürter, F. Reinbold, and W. Lorenz, A dipole antenna for interstitial microwave hyperthermia, IEEE Trans. Microwave Theory Tech., vol. 39, pp , June 1991.

7 88 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 1, JANUARY 2003 [8] S. Pisa, M. Cavagnaro, P. Bernardi, and J. C. Lin, A 915-MHz antenna for microwave thermal ablation treatment: Physical design, computer modeling and experimental measurement, IEEE Trans. Biomed Eng, vol. 48, pp , May [9] R. F. Harrington, Field Computation by Moment Methods, 1st ed. New York: McMillan, [10] I. Longo, C.N.R. Industrial Patent N. PI/2001/A/ ,. [11] Micro-Coax Components,, Collegeville, PA, Catalogue [12] Feko v Stellenbosch, South Africa: Electromagnetic Software of System SA (Pty) Ltd. [13] Z. Kauuk, A. Khebir, and P. Savard, A finite-element model of a microwave catheter for cardiac ablation, IEEE Trans. Microwave Theory Tech., vol. 44, pp , Oct [14] G. B. Gentili, M. Leoncini, and M. Vivoli, FDTD analysis of SAR distribution produced by interstitial hyperthermic applicators, J. Microwave Power Electromagn. Energy, vol. 29, no. 2, pp , [15] J. P. Berenger, A perfectly matched layer for the absorption of electromagnetic waves, J. Comp. Phys., vol. 114, no. 2, pp , [16] G. Mur, Absorbing boundary conditions for the finite-difference approximation of the time-domain electromagnetic field equations, IEEE Trans. Electromagn. Compatibility, vol. EMC-23, pp , Apr [17] D. B. Davidson, I. P. Theron, U. Jakobus, F. M. Landstorfer, F. J. C. Meyer, J. Mostert, and J. J. Van Tonder, Recent Progress on the Antenna Simulation Program FEKO COSMIG (Incorporating AP/MTT 98), Cape Town, South Africa, 1998, pp [18] E. Tombari, Private Communication, Istituto per i Processi Chimico- Fisici (IPCF),CNR, Pisa, Italy. [19] M. Kikuchi, Y. Amemiya, S. Egawa, Y. Onoyama, H. Kato, H. Kanai, Y. Saito, I. Tsukiyama, M. Hiraoka, S. Mizushina, T. Yamashita, Y. Nikawa, J. Matsuda, and M. Miyakawa, Guide to the use of hyperthermia equipment. 2. Microwave heating, Int. J. Hypertherm., vol. 9, no. 3, pp , Guido Biffi Gentili was born in Lucca, Italy, on August 9, He received the Ph.D. degree in electronic engineering in 1970 from the University of Pisa, Italy, where he also held a research assistantship in the field of radar system modeling and pulse compression. He is currently full Professor of Electromagnetic Theory and Techniques at the University of Florence, Florence, Italy, and is responsible of numerous research projects in the field of microwave engineering committed by private companies. His research interests are in the area of biological applications of microwaves, active and passive antennas and sensors, millimeter wave circuits, and microwave industrial applications. Matteo Cerretelli was born in Florence, Italy, on May 31, He gratuated in electronic engineering at the University of Florence, Italy, in In 2000, he joined the Department of Electronic and Telecommunications, University of Florence. His main research interests are in the analysis and design of automotive antennas, microwave sensors, and biological applications of microwaves. Iginio Longo was born in Lucca, Italy, on November 5, He received the B.S. in physics, from the University of Pisa, Pisa, Italy, in 1968.He received the First Researcher Degree from the National Research Council (CNR), Pisa, Italy, in 1988 Since 1970, he has been with the CNR, at the Institute for Atomic and Molecular Physics (IFAM), recently converged into the Institute for Chemical-Physical Processes (IPCF). His research field embraces atomic and molecular spectroscopy and quantum electronics. Currently, he is involved in the study of new spectroscopic techniques employing dielectric waveguides and resonators at microwave frequencies. (CNR), Pisa, Italy. Nevio Tosoratti was born in Rome, Italy, on April 1, He graduated with honors in electronic engineering in 1998 at the University of Rome La Sapienza, where he also received the Ph.D. degree in applied electromagnetism, developing broadband impedance measurement techniques on High T superconducting films in the RF and microwave field. At present, he is with the microwave spectroscopy laboratory at the Institute for Chemical-Physical Processes (IPCF) of the National Research Council