A Study on Application of Dielectric Resonator Antenna in Implantable Medical Devices

Worldwide, a large number of patients are benefited every year due to technological advancement in implantable medical devices (IMDs) such as in hyperthermia and bio-telemetry. The combination of sensors and antennas defined the quality of the implantable device. Antenna communities strive hard to fulfill the needs of the medical world by introducing new designs and concepts in this field. The purpose of this study is to identify major existing challenges and provide suitable solution of these challenges for implant applications. Present implant antennas have faced inferior performance due to high metallic losses and varied implant depth. The human body is a combination of dielectric materials, regardless of whether it is liquid (eg. water), gels or hard bones. In this study, first time a dielectric resonator antenna (DRA) resonating at 2.45 GHz, has been proposed as an implantable antenna with no metallic losses, varied implant depth performance and bio-compatibility. The implant DRA is placed on a bio-compatible polyvinyl chloride (PVC) substrate with a thickness of 0.5 mm. The rectangular DRA excited by a coplanar waveguide feed is proposed and its performance is compared with the help of four phantoms given in the literature. A detailed link design study was also undertaken in view of the different applications.


I. INTRODUCTION
A CTIVE Implantable Medical Devices (IMD) such as cochlear implants, haemodynamic support, implantable cardiac pacemakers, implantable defibrillators and implantable glucose monitors are procured by ambulatory centers, cardiac centers, dental clinics and hospitals for their patients. These are important devices that are responsible for biotelemetry: biomedical communication, transfer of physiological information including hyperthermia, blood details, cardiac beat and other such data which can be further used for timely treatment of present or future diseases in the body. Sensors and radiators are essential components of any IMD. In rare cases, the antenna can also act as a sensor because the quality of the signal strength at a fixed link distance can convey information about physiological data such as tissue density in cancer disease. Designing an implant antenna faces multiple challenges such as, miniaturization, efficient link budget, frequency detuning, corrosion, and displacement due to the aging of the body and the changing environment around the transmitter. Although initial research is more focused on the simplest single skin model, deeply implanted antennas considering fat and muscle are also needed. It is obvious that a deeper implant antenna will face greater challenges owing to the complex structure and composition of muscles in different parts of the body. In recent times the application of IMDs has increased significantly because of their life support. Research on wireless IMDs and other implantable sensors has increased over the last decade. The requirement of an efficient wireless communication link between On/Off body devices leads to the challenge of designing an optimum implantable antenna. The shift from the medical implant communication service (MICS) band (402-405 MHz) to industrial, scientific and medical (ISM) band (902-928 MHz and 2400-2483.5 MHz) is understandable due to the higher gain and possible miniaturization of implantable antennas. Different design considerations regarding implantable antennas have been discussed in [2]- [9] . Ingestible antennas [10]- [14] and ISM band implantable antennas [15]- [25] have also been discussed for diagnostics and monitoring purposes. These implantable antennas also strengthen the concept of 'doctor at home' as illustrated in Fig. 1, wherein patient can do 'work from home', while doctor can examine the reports sent by the 'human arm implanted transmitter' through the base station available at home.

A. MOTIVATION
Microstrip patch antennas are currently only solution for ingestible and implantable body antennas because of their various advantages, such as low profile and easy miniaturization. The human body is a combination of dielectric and conductive materials as shown in Table-1 [1]. Therefore, designing a metallic patch antenna for this lossy environment may become a tedious task but using a dielectric as a resonator in this high-permittivity environment may be a better choice. In the future, it may be possible that the human body which is a dielectric body, can act as a resonator as well as a radiator with all safety measures. Owing to the absence of metallic losses and high gain with design flexibility, DRA overcomes the disadvantages of microstrip patch antennas. Miniaturization is also possible in DRA if high permittivity biocompatible dielectric material like T iO 2 , Al 2 O 3 are chosen. In addition, in the case of a semiartificial limb replacement surgical scenario, implantable DRA can easily be customized according to the struc-tural conditions. In this study, a futuristic implantable dielectric resonator antenna is proposed. To the author's knowledge, no other implantable DRA work has been reported in the literature. In this study, the design steps for implantable DRA and a simple rectangular DRA (RDRA) with coplanar waveguide (CPW) feed are presented. For an in-depth study, four phantoms were used to demonstrate the performance of the proposed RDRA. These phantoms are rectangular single-layer skin models, a rectangular three-layer (skin-fat-muscle) model, improved cylindrical three-layer model and a 3D canonical right arm tissue model in ANSYS HFSS ® EM simulator. To strengthen the study bio-compatibility and specific absorption rate analyses have also been provided. A link design study is also a very important part of the implanted antennas. An efficient path loss model for in-body to off-body communications in the ISM band was discussed in [26]. Various biomedical applications differ on the basis of data rate, transmitter power and link distance for example, saturation of peripheral oxygen (SpO 2 ), blood pressure or electrocardiogram (ECG) may require a data rate as low as 100 Kbps and on the contrary, multi-channel electroencephalogram (EEG) sensors or capsule endoscopes may require a 1.5 Mbps data rate [27]. For any biomedical application, a significant data rate and link distance are prominent characteristics to ensure that discriminative information is delivered within an acceptable time frame for the most critical scenario. This study demonstrates that DRA can be used in biomedical applications. In the future we may see 'human body-like material' based 'gel-DRA' which may easily adapt and perform harmlessly and efficiently in the lossy and harsh environment of the human body. The remainder of this paper is organized as follows: Introduction section discusses the need and motivation for this study; the second section explains the manual and algorithm based methodology to design an implantable DRA and provides insight into biocompatibility issues in implant antennas; the next section discusses radiation mechanism in four different phantoms and study implant DRA with the help of simulated result analysis including evaluation of radiation pattern, gain, impedance bandwidth, and specific absorption rate; the next section briefly discusses the link budget analysis for the proposed DRA; followed by conclusions and references. Implantable antennas have a wide range of applications such as IMDs (pacemaker, defibrillators etc.), ingestible capsules (for endoscopy, intestinal diagnostics etc.), hyperthermia, cancer detection and other biotelemetry applications. After aiming for the application and operating frequency (f 0 ), a DRA can be designed considering the following points: 1. DRA height and substrate thickness should be minimum to achieve miniaturization,; 2. Biocompatible materials may be used for DRA, and substrate or biocompatible material coating may be used on the substrate and DRA as the antennas are to be implanted inside the body,; 3. Excitation methodology, such as probe feed, microstrip line feed, aperture couple feed, and coplanar waveguide feed, must be chosen, considering the internal body environment and maximum power limitation with respect of RF exposure and specific absorption rate values as per international standards.  [32], [33], [35]- [38]. Biocompatible blends based on polyvinylchloride (PVC) and natural polymers can also be used in medical device fabrication [33]. Mostly, DRA are made up of biocompatible ceramics which are widely used in Periodontology and Implant Dentistry [39], [40]. These ceramics have already proven safe for human body. In this study a cuboid DRA made up of ceramic of T iO 2 with ϵ r =80 is placed on PVC plastic substrate with ϵ r =2.7 of thickness 0.5 mm. The approximate dimensions of the DRA were calculated using Marcatili's approximation method [46]- [48] at 2.45 GHz for the T E 11δ mode given by Eq. (1)(2)(3)(4).
where, k x , k y , k z denotes wavenumbers along x, y, z directions, respectively. k o denotes free space wavenumber corresponding to resonant frequency f o and a, b, d are length, breadth and height of rectangular DRA for relative dielectric constant range of 10<ϵ r <100. To avoid the ground plane at the bottom side of the substrate which can lead to losses and high power requirements to achieve desired results inside the body environment, a coplanar waveguide feed is used to excite RDRA. In implant antennas, the dimensions of the device play an important role; therefore thickness of the substrate is fixed at 0.5 mm and the height of the RDRA is also locked at 5 mm. The dimension of DRA are compact 23.5 mm x 23.5mm x 5 mm which can be accomodated in implantable devices easily. It is made up of biocompatible ceramic so it weigh less which made it suitable for implant applications. The overall thickness of the DRA and substrate is 5.5 mm which is much less than that of previously published implantable antennas with physical dimensions of (62× 35× 7.8 mm 3 ) and (54× 31× 11 mm 3 ) in [49] and [50], respectively. Hence, we believe that physical size of the discussed implantable DRA is fit for limb muscle implantation. In Fig. 2 and Fig. 3 two design methodology are shown to design the DRA for implantable applications. These are manually parametery-based optimization and advanced algorithm-based optimization of DRA. There are different advanced optimization algorithms that are responsible for minimizing the cost function provided to them. These are genetic algoritm (GA), quasi-newton, pattern search, sequential gradient, different variants of VOLUME 4, 2016 particle swarm optimization (PSO). Both the methods are efficient but in our case algorithm-based optimization consumes more time, as it took approximately 2 h 45 min to complete one iteration in the 3D canonical right arm tissue model in ANSYS HFSS ® Electromagnetic simulator using a computer with an Intel ® i7 (2.4 GHz) processor, 16 GB RAM and 4 GB graphics card. Therefore, the manual optimization methodology mentioned in Fig. 2 has been adopted here to achieve optimized dimensions of DRA. Four phantoms were used in this study: 1. The rectangular single-layer skin model (M-1) mentioned in [2], [7], [11], [17]- [19] with d1=30 mm and d2=100 mm; 2. The rectangular three-layer (skin-fat-muscle) model (M-2) mentioned in [2], [7], [11], [17]- [19] with f3=12 mm and f6=100 mm, fat and skin thickness 4 mm each,; 3. The cylindrical three-layer (skin-fat-muscle) model

B. SPECIFIC ABSORPTION RATE (SAR)
Electromagnetic waves emitted by wireless devices should not cause medical concerns for users. To asses potential risks from electromagnetic waves, a safety parameter known as SAR is adopted by international agencies which is defined as "The time derivative of the incremental energy (dW ) absorbed by (dissipated in) an incremental mass (dm) contained in a volume  [28]. The IEEE C95.1-1999 standard restricts the SAR averaged over any 1 g of tissue to be less than 1.6 W/kg [28], and the IEEE C95.1-2005 SAR averaged over any 10 g of the tissue be less than 2 W/kg [29].Therefore, standard C95.1-1999 is stricter than standard C95.  Table-3. The maximum value of the SAR field is 388.38 W/kg and 53.97 W/Kg for 1-g avg and 10-g avg cases respectively, and maximum allowed net input power of both cases, calculated by simulated implant antenna model in human arm phantom by decreasing the input power to reach the international standard limits, are given in Table-3. It is confirmed from tabulated data that maximum allowed net input power for both the standard are 2.25 mW and 37 mW which are more than required when input power is less than 1 mW provided by internal battery.  [3], [34]. Zirconia is preferred by designers because of its high permittivity and low loss tangent which allow the near field of the implanted radiator to concentrate inside the encapsulation layers and to reduce power loss [35].

III. DRA RADIATION MECHANISM AND RESULT ANALYSIS
In this study, the behavior and patterns of implanted DRA inside the human body were studied by using an electromagnetic simulator have been studied. A rectangular single-layer skin model as shown in Fig. 5, was adopted for implanting the DRA. It is a well established fact that the human body is a lossy dielectric one, so it was interesting to see the S-parameter variation of a given DRA in a rectangular single-layer skin model comprising skin tissues with electrical parameters, as shown in Table 1. The depth of the implanted DRA inside this model has been varied in practical permissible limits (depending upon human body anatomy) and its effect is shown in Fig. 7. This indicates that as the depth increases, impedance matching deteriorates which indicates a trade-off between both. However, in this case orientation fluctuation was ignored to reach a conclusion. It can be seen that the 'depth of implanted DRA' introduces a gap of 'moderate dielectric constant medium' called skin which provides a smooth transition of dielectric medium, from high-permittivity (DRA) to low-permittivity medium (air). Because this inserted medium is also lossy with high tan δ therefore it attenuates the radiation. Therefore, the DRA may show proper impedance matching in this 'depth window' (depth=2. 5-4.5 mm) where tradeoff between smooth transition and attenuation, both provided by the skin layer hold in favor. In [51], Snell's law was used to describe the radiation mechanism of the implanted antenna which also holds good here, because varied dielectric mediums involved in the body and electromagnetic waves undergo refraction under these conditions which causes the spread of electromagnetic energy in a denser medium and hence attenuation of wave. Figure 9 shows the simulated gain variation at 2.45 GHz, ϕ = 0, for different depths of the implanted DRA in the single-layer model. The gain of the implanted antenna in the broadside direction (at θ = 0 and ϕ = 0) varies from -36 dB to -29 dB moving deeper inside the skin in the suitable 'depth window' defined earlier.  In the literature, three-layer model with skin-fatmuscle was used to justify the performance of the antenna in approximate body conditions by including three human tissues as shown in Fig. 5 [2], [11], [17], [18]. TheS 11 parameters for different implant antenna depths inside three-layer model are shown in Fig. 10. It represents that as implant antenna depth increases, impedance matching deteriorates, similar to singlelayer model case. But here the useful 'depth window' (depth=4.5-10 mm) is fairly large. Figure 11 shows the interpretation of the radiation mechanism in a threelayer model. This indicates that as the depth of the implanted antenna increases, the dielectric constant of different media causes abrupt transition which results a loss of signal and poor impedance matching. Figure  12 shows the simulated gain variation at 2.45 GHz, ϕ = 0 for different implant antenna depths. It depicts that the gain in the broadside direction varies from -14.9 dB to -25 dB for the 'depth window' in the threelayer model case. It is also observed from Fig. 12 that, as the depth increases radiation from transmission line dominates more due to impedance mismatch.
The two previous phantom models discussed were rectangular in nature whereas the human arm was round shaped. Therefore, to increase similarities with the natural human arm, a rectangular three-layer model was modified to a cylindrical shape, keeping all three layers (skin-fat-muscle) along with bone, as shown in Fig. 5. Table-1 Fig. 14. At depth=4.5 mm best impedance matching is observed throughout the operating band, and the simulated gain is -14.02 dB. It is also observed from Fig. 14 that, back radiation from the transmission line dominates more due to impedance mismatch, as the depth of the implant increases. Figure 15 shows radiation mechanism in the cylindrical model, and it is observed that as the depth decreases, the curvature effect of the cylindrical model becomes stronger. Table-4 shows the electric field distribution in the cylindrical model for the four implant depths. This shows that electric field is trapped between two high-permittivity mediums, that is skin and muscle, which results in the loss of radiation. It is also essential to mention that effective wavelength (λ ef f ) varies with the medium and λ ef f for skin, fat and muscle is calculated as 19.85 mm, 53.26 mm and 16.85 mm, respectively, from λ ef f = λ 0 / √ ϵ r where λ 0 is wavelength in air or vacuum. Similar to the previous models (M-1 and M-2), here, a 'depth window' (depth=4 mm-7 mm) is found in which the implant antenna  Finally, a simulation analysis was performed on the 3D canonical right arm tissue model using commercially available full-wave FEM based Ansys HFSS (v19.3) software in the frequency domain. Far-field distance for the proposed antenna at 2.45 GHz frequency is approximately 25 mm (equivalent to λ 0 /5). The antenna characteristics have been measured at a distance of 50 cm from the skin of human arm model which lies in far-field region. Due to smaller mesh size for more accurate results complete human body with anatomical arm could not be simulated in an available 16 GB RAM, i7 processor (2.4 GHz) machine, which force us to take smaller body part. So, the upper human arm was chosen because it has thick muscle and fat, which resembles the chest and thighs, but the optimization time taken by the EM solver effectively reduced as compared to  Fig. 16 and Fig. 17 for different implant antenna depths. Owing to the asymmetric construction of the human body and variable dielectric constant, the S 11 parameter varies instantly for different depths, but it makes a 'depth window' (depth=2.3-5 mm) for which the impedance bandwidth is under -10 dB in the band of interest. As expected gain drops and varies from -23.6 dBi to -29.5 dBi for different implant antenna depth in the 'depth window' in broadside direction (here at θ = 90 and ϕ = 0) due to change in inclination of human arm model in the HFSS solver. To verify the results obtained from HFSS solver which is based on finite VOLUME 4, 2016  In this section, a generalized link budget analysis is undertaken in view of different applications for which implanted antennas have been proposed. EIRP, or effective isotropically radiated power, or equivalent isotropically radiated power, is a measurement of output power in one direction from an ideal, one-dimensional source. The ideal point-source radiates electromagnetic energy in a spherical pattern, so the "maximum" radiated power should be the same in all directions. In link design theory, the EIRP is represented by Eq. (5).
P t , G t , and L f eed are transmitter power, the transmitter antenna gain, and feeding loss, respectively . Dipole antennas are assumed to be well matched, so their impedance mismatch losses are negligible. Polarization mismatch losses were also neglected. The impedance mismatch loss is represented by Eq. (6) and for S 11 =-10 dB, -15 dB, -20 dB, the impedance mismatch losses are calculated as 0.458 dB, 0.14 dB and 0.044 dB, respectively, which can be neglected in link design.
L imp (dB) = −10 log 10 ( where |Γ| is magnitude of the reflection coefficient. Free space path loss is an important factor in link design and is represented by Eq. (7). Here, the implant antenna depth inside the human body is neglected compared to the distance of the external receiver antenna. Therefore, the medium between the implanted antenna and receiver antenna is assumed to be free space. Therefore, it is important to consider the factor because the wireless independence and power of the transmitter depend primarily on the optimum distance between the links. For a body worn or nearby located external antenna in the room of the patient, distance may be compromised for a lower value, say d < 5 m.
At the receiver, the noise power density N 0 depends on the noise factor (N F ) as shown in Eq. (8) and Eq. (9), this N F is taken as 2.5 or 3.5 depending upon the dipole antenna used. The ambient temperature (T 0 ) of the receiver also affects the value of the noise power density which is generally taken as room temperature (293 K).
To design a feasible link, the margin between the required C R /N 0 and the current link C/N 0 should be non-negative. Link C/N 0 , required C R /N 0 and link margin (LM) are described by Eq. (10), Eq. (11) and Eq. (12), respectively. -4 78X10 3 [17], [25] - 40 7 Equation (13) shows that LM depends heavily on the transmitter power (P t ), bit data rate (B r ) and distance (d) between the transmitter and receiver (in free-space path loss). The transmitter power largely depends on the allowable maximum power limits in accordance with the SAR standards. The bit data rate depends on the biomedical application; for example, blood pressure, heart rate, blood oxygen and glucose monitoring require a maximum data rate of 40-200 samples/second. In biomedical applications, higher data rates are preferable to reduce latency and ensure timely delivery of critical biomedical data to health professionals. However, higher data rates can be sacrificed in some applications to manage power efficiency at larger link distances.

A. LINK BUDGET PARAMETERS FOR PROPOSED RECTANGULAR DRA
In this section, only one case of the 3D human arm model is used to calculate the link budget parameter of proposed RDRA out of the number of cases and models discussed. However, in biomedical applications, different link budget parameters can also be calculated. Using Eq. (5)(6)(7)(8)(9)(10)(11)(12), the link design parameters were calculated for 25 µW (-16 dBm) transmitter power and are listed in Table 6. In this case, LM was found to be positive, which emphasizes the feasibility of the link setup for discussed implant DRA. It has been observed that researchers used different sets of transmitter power and bit data rates to transfer the signal from the transmitter to the receiver, in the implant antenna wireless link design, as shown in Table 5.
To better understand the effect of the transmit power and bit data rate on the performance of the antenna, various values of LM have been calculated for different link distances, transmit powers and bit data rates given in Table-5. Figure 19 and 20 show the variation of LM for the best (-23.6 dB) and worst (-29.5 dB) antenna gains in the 3D human right arm case. It can be observed from Fig. 19 and 20, that, the proposed RDRA can be utilized for different biomedical applications under ubiquitous circumstances.

V. CONCLUSIONS
In this dissemination, first time design steps and studies on implantable DRA have been presented. The idea of using DRA in biomedical applications was crafted using a rectangular DRA mounted on a PVC substrate. Two design steps were used to construct the implant DRA with dimension 23.5 mm x 23.5 mm x 5 mm which makes it light and compact to be used in implantable devices. The performance of the proposed DRA was studied in a rectangular single-layer skin phantom, rectangular three-layer (skin-fat-muscle) phantom, cylindrical threelayer (skin-fat-muscle) phantom, and 3D canonical right arm tissue model in the HFSS EM simulator. In all four body models, the proposed DRA shows acceptable VOLUME 4, 2016 LADISLAU MATEKOVITS received the degree in electronic engineering from the Institutul Politehnic din Bucureti, Bucureti, Romania, and the Ph.D. degree (Dottorato di Ricerca) in electronic engineering from the Politecnico di Torino, Torino, Italy, in 1992 and 1995, respectively. Since 1995, he has been with the Department of Electronics and Telecommunications, Politecnico di Torino, first with a post-doctoral fellowship, then as a Research Assistant. He joined the same Department as