MICROWAVE BREAST imaging has been proposed to. A New Breast Phantom with a Durable Skin Layer for Microwave Breast Imaging

Transcription

1 A New Breast Phantom with a Durable Skin Layer for Microwave Breast Imaging John Garrett, Student Member, IEEE, and Elise Fear, Senior Member, IEEE, Abstract Breast phantoms are required to test and validate microwave breast imaging prototypes. For this purpose, a new breast phantom made from carbon/rubber mixtures is proposed. These materials have (a) electrical properties that are stable over time and representative of human target values, and (b) mechanical properties that allow the material to be flexible and withstand reasonable stress. To characterize and optimize the carbon/rubber materials, samples made with varying carbon concentrations were created and the dielectric properties were measured. From these materials, a skin layer, fatty layer, glandular structures and phantom tumors were cast from 3D printed molds and assembled into a complete breast phantom. These phantoms mimic the anatomical structures of the breast, are reconfigurable for a variety of tests, and are easy to create in a typical lab environment. A microwave breast imaging prototype system was used to measure reflections from breast phantoms. Comparison with reflections from human trials demonstrated that the phantom provides appropriate skin reflections. Phantoms incorporating glandular structures were imaged using a delayand-sum technique. A response consistent with the position of the inclusions was observed. Overall, the carbon-based phantoms provide similar reflections to human tissue, and have proven useful for testing our imaging algorithms. Index Terms Dielectric materials, Dielectric measurement, Biomedical engineering, Biomedical imaging I. INTRODUCTION MICROWAVE BREAST imaging has been proposed to monitor breast health and several prototype systems have been developed [] [7]. This imaging modality measures microwave signals reflected by and/or transmitted through the breast to reconstruct backscatter intensity or dielectric property images of its internal tissue structure. This can be used to monitor breast tissue health using the observable dielectric property (DP) contrast that exists between healthy and malignant breast tissue [8]. Successful translation of these prototypes into clinical settings requires system validation and testing. To address this issue, breast phantoms (BPs) with realistic breast tissue structure and DP characteristics are needed. The DPs of biological materials are directly related to tissue water content [9]. This has lead to the abundant use of water-based BPs tailored to mimic physiological breast tissue. Initial saline-based breast phantom materials (BPMs) incorporating polyethylene powder and a gelling agent [] were adopted for ultrawideband applications using a gelatinoil mixture []. BPs made from materials similar to [] and [] have been created or proposed, ranging from simple heterogeneous structures [] [6] to more complex designs including skin, fat, glandular and tumorous tissue regions [7], [7] []. Additional water-based BPs using flour [] or glycerin [], [3] mixtures contained within rigid, plastic breast shaped molds have been reported. Although realistic DPs and structures have been created, there are several disadvantages to using water-based BPs. Firstly, many of these BPs were developed for tomographic systems where a realistic skin is not as important for validation purposes as it is for radar-based systems. Some of these BPs do feature realistic skin layers [7], [4], [7] [], [4]; however, many either do not have a skin layer or the skin layers have unrealistic properties [], [3], [6], [] [3]. Secondly, most of these BPs are not mechanically robust and require hard plastic shells to maintain their shape [5], [6], [] [3]. Shells are likely needed for testing in imaging systems where the BP must support itself, i.e., imaging systems where the breast is not supported by a dielectric dome. Although this shell should not present electromagnetic contrast with respect to the immersion medium, it does prevent the compression of these BPs, which is required by some systems (e.g. [5]). Thirdly, water evaporation in the water-based BPMs leads to changing DPs; therefore, BPs typically degrade over time and are required to be cooled and covered in plastic during storage in order to prolong their usefulness. These storage requirements can be inconvenient because the BP must be allowed to warm up to room temperature before imaging. The reported stability of these BPs ranges from 9 weeks [7] to just 5 days [4]. Alternatively, BPMs have also been made from mixtures of carbon powder and an insulating matrix. For the insulating matrix, ceramic powder [6], [7], plastic resin [8], silicone rubber [9], [3] and urethane rubber [3] have been tested. For applications such as BPs, silicone and urethane rubber are ideal due to the flexibility of the material. A carbon/urethane mixture has previously been shown to possess many of the electrical and mechanical properties required to create a BP [3]. An ideal BP has (a) DPs that are close to target human values and stable over time, (b) good mechanical properties so that it can withstand reasonable mechanical strain and flex under compression, (c) a construction that approximates the anatomical structure of tissue found in human breasts and provides predictable dimensions, (d) the ability to reconfigure the structure for a variety of tests, and (e) low-cost materials that do not require expensive equipment. Based on these criteria, a new BP has been created. This work has been divided into 3 sections for this paper. First, the development of the BPMs is outlined in section II-A with the DPs presented in section II-B. Second, the complete BPs are presented with appropriate DPs, skin layers, fat layers and inner glandular structures in section III. Finally, reflection Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

2 TABLE I: CARBON POWDERS AND RUBBER MATERIALS USED DURING BPM DEVELOPMENT. TABLE II: CHOSEN CARBON-BASED MIXTURES FOR VARI- OUS SOFT TISSUES FOUND IN THE BREAST. Material Supplier Location Carbon black, 5% compressed Alfa-Aesar Ward Hill, MA Graphite, powder, < µm Sigma Aldrich St. Louis, MO Urethane rubber, PMC R /3 Smooth-On Easton, PA measurements of the BP using a monostatic radar imaging prototype are presented in section IV along with backscatter energy maps created from these reflections. II. BREAST PHANTOM MATERIAL DEVELOPMENT A. Methodology Due to the requirements of an ideal BP, a mixture of carbon-based conductive filler and rubber matrix was chosen to create the BPM. Urethane was selected for the rubber matrix because it possesses superior mechanical properties with respect to silicone rubber as explained in [3]. For the conductive filler, graphite and carbon black were tested. Specific material information is given in Table I. Different concentrations of carbon powder were tested to create materials with a range of DPs. In previous work, it was found that graphite/urethane mixtures did not provide adequate DPs, and carbon black/urethane mixtures provided poor mechanical properties [3]. Carbon black/graphite/rubber mixtures were able to simultaneously provide good dielectric and mechanical properties. A range of concentrations were tested in this previous study; however, ideal concentrations for BPs were not identified. To create the BPM samples, urethane solution (prior to curing) and carbon powder were weighed, poured into a container, and mixed by hand with a metal stirrer for several minutes. The mixing was done in a fume hood due to the toxicity of carbon black in the powder form. After the BPM mixtures had set, a dielectric probe [3] was used to measure the DPs in three locations from to GHz. For each combination of materials tested, cylindrical samples with 5 cm diameters and 3 cm heights were created. This permitted the collection of multiple measurements without overlapping the sensed volumes. The average of these measurements is used for plotting the results shown in section II-B. In order to properly mimic human tissue, these materials should provide DPs that are similar to human tissue. Similar to previous studies, experimental measurements of tissues are included in section II-B for comparison. This includes dry skin [33], -3% adipose (high-property glandular), 3-84% adipose (low-property glandular) and 85-% adipose (fatty) tissue measurements [9]. The mechanical properties of the BPMs were also assessed to determine the elasticity and the tensile breaking point. For this test, tensile force was applied to samples (i.e. tendons ) with a uniaxial testing machine (HKT, Tinius Olsen, Rock Hill, SC) until they tore apart. Tissue Type Graphite Carbon Black Urethane (wt%) (wt%) (wt%) Fatty Skin Glandular Tumor B. Results and Discussion The results from [3] were used as a starting point, but changes in carbon concentration were required to optimize the BPMs for use in BPs. The chosen material concentrations for skin, glandular and fatty BPMs are given in Table II, and the DP results are shown in Figure along with experimental measurements of breast tissues for comparison. As seen in Figure, the properties of the BPMs are lower than the target human values. The skin BPM required a lower carbon concentration than the glandular BPM due to the mechanical property requirements of the skin layer. Although the properties of the chosen skin BPM are lower than the target human values, it may still provide reflections similar to human skin. This will be investigated in section III-B. Relative Permittivity () Conductivity (S/m) Glandular BPM Skin BPM Fat BPM High Glandular Low Glandular Fat Skin Fig. : DPs of BPMs optimized for use in BPs with dry skin [33], high-property glandular, low-property glandular and fatty tissue [9] DPs included for comparison. The tensile strength of the skin BPM was measured to assess the mechanical properties. Three samples were measured, and an average was found. The skin BPM had an average ultimate stress of.6 MPa and an average ultimate elongation of 8%. With this skin BPM, BP skin layers can deform under compression, but the shape of the phantom structures is still maintained under the weight of the materials. Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

3 3 A. Methodology III. BREAST PHANTOM DEVELOPMENT A realistic BP should include the same structures that would be present in either a healthy or cancerous human breast. This includes skin tissue, fatty tissue, glandular tissue, and possibly tumor tissue. The skin layer was chosen to be mm thick, cm in diameter and 9 cm deep (from chest wall to nipple), which is approximately anatomically correct []. A mold was therefore required to ensure consistent size and thickness of the skin. For this, a mold consisting of an outer shell and an inner shell was created with a 3D printer (Replicator, MakerBot Industries, Brooklyn, NY). To create the skin layer, the outer shell was filled with the carbon/rubber mixture and the inner shell was pressed down. As the inner shell was pressed down, the mixture spread from the BP s nipple to the BP s chest wall. The outer shell was printed in two halves to facilitate the removal of the skin. The mold is shown with one half of the outer shell removed in Figure. Internal structures were also created with 3D printed molds to roughly mimic the shape of glandular tissue. By using 3D printed molds, many different shapes can be created for the glandular tissue inserts. By controlling the concentration of carbon, it was possible to create a range of DPs for these internal structures. The carbon/urethane rubber was also very easy to cut into abstract shapes; however, for initial testing, simple shapes were chosen. The internal BP structures were glued to nylon threaded rods and bolted to a polycarbonate disc to control their position inside the skin layer. The polycarbonate disk had many different holes so many different configurations could be created by rearranging the glandular inserts. The internal structure assembly was then placed inside the skin layer. An example configuration is shown in Figure 4. Fig. 4: An example of an internal structure configuration attached to a polycarbonate disk (right) that is ready to be placed inside a skin layer (left). Two possible BP tumors attached to polycarbonate rods are shown below. Fig. : The 3D printed mold used to create the BP s skin layer. The outer shell is shown with one half removed. Figure 3 displays two skin layers created using this process. The model on the left has vertical sidewalls; however, this shape became difficult to create when using thick mixtures during the molding process. The model on the right was created to allow the carbon/rubber mixture to flow towards the chest wall more easily during the molding process. BP tumors were also created with a similar process. These phantom tumors were made with higher carbon concentrations than the phantom glandular tissues and cut to various sizes. The phantom tumors were then placed on 3 mm diameter polycarbonate rods and attached to the polycarbonate disks to secure their position inside the skin layer. Two phantom tumors are visible in Figure 4. The interior of the skin was filled with material to mimic the DPs of fatty tissue. For this, a carbon/rubber mixture could be used (such as the one listed in Table II), but this would not allow the configuration to changed once the rubber has set. Alternatively, since fatty tissue has very low DPs, canola oil or glycerin can also be used to mimic fatty tissue, which allows the BP to be easily reconfigured. Fig. 3: Two BP skin layers: one with vertical side walls (left) and one with 8 side walls (right). B. Results and Discussion The resultant skin layers had very few air bubbles. The bubbles that were present were easily cut with a knife and filled with additional carbon/rubber mixture. Each skin layer was also checked for thickness with a set of digital calipers. The skin on the right in Figure 3 was measured in different locations and had an average thickness of. ±.8 mm. For the fatty tissue, several materials were mentioned in section III-A. The DPs of these fatty BPMs were measured Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

4 4 using a dielectric probe [3], and the dielectric properties are shown in Figure 5. Relative Permittivity () % Graphite Glycerin Canola Fatty Tissue Conductivity (S/m) Fig. 6: A diagram of the monostatic radar imaging prototype including a diagram of a breast for illustration Fig. 5: DPs of materials that are available to mimic fatty tissue including the target human value [9]. With the chosen glandular configuration fixed to the polycarbonate disc, the polycarbonate disk was bolted to a plastic ring, clamping the phantom skin in place (as seen in Figures 3 and 4). This technique allows for a variety of BPs to be created ranging from simple homogeneous breast models to complex breast models with heterogeneous interiors. When sealed, the BPs were able to contain a liquid (e.g. water) for several days without leaking and withstand moderate compression without the skin deforming or tearing. Combined with the tensile strength results in section II-B, this shows that the BPs are mechanically durable and suitable for repeated testing. IV. BREAST PHANTOM TESTING The final BP described in section III has both realistic dielectric properties and structures. Next, the suitability of these phantoms for imaging was assessed by analyzing reflection measurements taken with a monostatic radar prototype. A. Measurement Methodology The monostatic radar prototype features an ultra-wideband (UWB) antenna [34] that is adaptively positioned relative to the object of interest [35]. As shown in Figure 6, the antenna and BP were submerged in a tank that was filled with canola oil. The antenna was placed cm from and perpendicular to the surface of the BP. To achieve this, 4 degrees of freedom were used for antenna movement. In Figure 6, two degrees of freedom are indicated. The pitch and radial location of the antenna may also be adjusted. When measurements were collected, a laser (located on the arm to which the antenna is attached) was used to acquire the profile of the BP. The profile was used to determine the locations and orientations of the antenna such that it was perpendicular to the surface of the BP. To scan the BP, measurements were taken at 4 locations. Specifically, the antenna was repositioned at 7 elevations at each of angles equally spaced around the perimeter of the tank. Once the antenna was repositioned at a measurement location, a vector network analyzer was used to collect the measurement data (PNA-X N54A or PNA-L N53A, Agilent Technologies, Santa Clara, CA). Measurements were acquired at frequencies between MHz and GHz with khz IF bandwidth. The average of 3 measurements was recorded. The frequency domain data was converted to a time-domain pulse by weighting the measured reflection coefficient with the spectrum of a differentiated Gaussian pulse and taking the inverse chirp z-transform [36]. The differentiated Gaussian pulse had maximum frequency content at 4 GHz and full-width half-maximum extending from.3 to 7.6 GHz. Finally, a calibration step was performed. Measurements were collected at the same set of antenna locations with an empty tank (i.e. BP removed). These measurements were subtracted from the measurements collected with the BP present in order to isolate the reflections from the BP. B. Results ) Consistency of Reflections: To assess the consistency of the skin layer structure and composition, reflection measurements of the BP were investigated. The BP used for testing had a skin layer composed of 3 wt% graphite and 4 wt% carbon black, and was created with 8 sloped sidewalls (shown on the right in Figure 3). The interior was filled with canola oil to mimic the fatty tissue inside the breast. Figure 7 shows measurements collected at a selected elevation (roughly 4 cm from the apex), including the data collected at each of the angles described in section IV-A. All reflections are very similar and provided an average correlation of.96, which suggests that the phantom skin layer had very consistent properties. As the antennas are a consistent distance away from the BP and illuminate approximately 4 mm by 3 mm (half-energy beam width at the skin), consistent reflections are expected with consistent skin layer thickness and properties. Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

5 5 Magnitude () Fig. 7: Skin reflection measurements from radial locations taken from a central location between the apex and lid of the BP. ) Representative Reflections: To determine whether the reflections collected from the BP are representative of reflections from human scans, the BP scans are compared with reflections collected from volunteers. The focus here is on the reflections from the skin layer. The signals of interest were collected during scans of two volunteers recruited into an ongoing patient study (study 498 as approved by the University of Calgary Conjoint Health Research Ethics Board). These scans were performed with the prototype system shown in Figure 6, however a patient bed was fitted to the top of the tank. During a scan, a volunteer lies on her stomach with one breast extending through a hole in the patient bed and into the tank filled with the immersion liquid of canola oil. Patient-specific antenna positioning was achieved using the laser profile approach described previously. The laser profile was also used to identify antenna locations approximately 4 cm from the apex of the breast. Data were converted to the time domain using the method described previously. For both BP and volunteer scans, reflections collected approximately 4 cm from the apex of the model or breast were identified. The reflections were then averaged separately in the time- and frequency-domains point-by-point. As seen in Figure 8, the response from the BP was very similar to the response from the participants breasts. This suggests that, although the BP skin may have lower DPs than reported in literature (e.g. [9]), skin reflections similar to human tissue are produced. It is, however, inaccurate to directly compare the BP and patient results since the shape of the breast affects the reflection and is expected to differ significantly between each volunteer and the BP. Overall, there will be a range of acceptable skin reflections because this reflection varies person-to-person; Figure 8 demonstrates that the phantom skin has an acceptable response. To further examine the impact of a skin layer with lower properties on the reflections from the phantom, simulated breast models were used. Simulations were created in FDTD electromagnetic simulation software (SEMCAD X, SPEAG, Switzerland), and featured an UWB antenna (the same antenna that was described in section IV-A), a BP and an immersion liquid of canola oil. The BP consisted of a mm thick skin layer (using the dimensions of the BP shown on the right in Figure 3) filled with canola oil. The UWB antenna was placed in a position consistent with the description in section IV-A and at a central elevation (i.e., between the lid Magnitude (db) Magnitude () 4 Participant # Left Participant # Right 6 Participant # Left Participant # Right Breast Phantom Fig. 8: Average skin reflection taken at 4 cm above the lowest extent of the sloped BP in the tank and from volunteers for comparison. and the apex of the BP). For the skin layer, two different DPs were simulated: the skin BPM DPs (Figure ) and DPs taken from literature [33]. Unlike Figure where dry skin was used as the skin value from literature, wet skin was used here to provide a worst-case scenario (both skin values are from [33]). The reflected signals are shown in Figure 9. The simulated reflections had a very high correlation coefficient of.9743, and as expected, the peak-to-peak magnitude of the simulated reflection from the skin layer using results from literature is about 3% higher. Again, the properties of human skin tissue will vary between individuals, and this skin BPM represents an acceptable response. Magnitude.5.5 x 3 DPs from Literature Skin BPM DPs Fig. 9: Simulated reflections from a BP skin layer using DPs taken from literature and DPs taken from measurements of the skin BPM. Additional simulations were created to determine the effect of the change in skin reflection on imaging. The same simulation model was used and again the two different skin layer DPs were simulated; however, a 6 mm diameter tumor was also included (ɛ r = 4, σ = S/m). These dimensions are based on the physical BP tumor, which is used section IV-B3. The tumor response was isolated by subtracting the reflection results simulated without the tumor present. The peak-to-peak reflection magnitude of the tumor was then compared to the total peak-to-peak reflection magnitude for the antenna closest to the tumor (the tumor is at a depth of cm below the Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

6 6 skin layer for this antenna). This gave peak-to-peak magnitude ratios of 9.% and 4.3% for the skin BPM DPs and the DPs found in literature, respectively. These results suggest that detecting the tumor in the BP is less challenging than in a model with more realistic skin properties; however, this will also depend on the accuracy of the skin subtraction algorithm and the sensitivity of the system because the reflection is small in both cases. The skin DPs from literature also represented a worst-case scenario, and the properties of human skin would likely be lower. 3) Imaging: Key aspects of the BP are the reconfigurable nature and predictable dimensions as a result of the manufacturing process. A reconfigurable BP allows a set of increasingly complex BPs to be created, giving insight into challenges related to imaging. By progressively increasing the complexity of the model, reflections from different components can be isolated. Predictable dimensions enable simulation models to be implemented for comparison with measurements, providing understanding of the impact of system performance on imaging. To illustrate these aspects of the model, several different variations on the BP were examined. The models consist of several components. The skin layer is the same skin that was used previously in Figures 7 and 8 (8 slope to sides, 3 wt% graphite and 4 wt% carbon black). Initially, the skin layer was filled with canola oil. Next, a 4-cm diameter cylinder extending from the tip of the breast to the polycarbonate lid was included. Finally, a tumor was placed near to the inclusion. Since all of the DPs and geometries of the BPs are well known, a simulation model representing the BP was created. Aside from the inclusions (i.e., the cylinder and tumor), this simulation was very similar to the simulation from section IV-B. Figure shows one of the simulated BPs. Fig. : Computer generated model of the BP with a 4 cm diameter inclusion. First, reflections from the BP were collected with only the skin layer and canola oil present. Then the glandular insert was added and the measurement process was repeated using the same antenna positions. From this data, a single antenna position was chosen to investigate the effect of the glandular insert. Figure a shows the difference between reflections collected with and without the gland present in the model. This measurement process was then simulated using Magnitude () Magnitude () Skin + Gland Skin (a) Skin + Gland Skin (b) Fig. : Reflection data from a single antenna position showing the response due to the glandular insert with (a) measured data and (b) simulated data. the model shown in Figure, and the results are shown in Figure b. These signal plots both show a dominant skin response however the glandular response is evident by comparing the reflection measurements taken with and without the glandular insert present. Together, Figures a and b show good agreement between simulations and measurements. The glandular response shown in Figure suggests that, if an imaging technique is used, an outline of the glandular structure will be reconstructed. To investigate this, we used a delay-and-sum technique to create a 3D image []. The measured response from the glandular structure was isolated by subtracting the response measured with only the skin present. The imaged region was defined by the surface of the model. The image was created by scanning the focal point through the imaged region, and summing the appropriately time-delayed signals recorded at each of the antenna locations. A central slice is shown in Figure a. Using the same image reconstruction technique, simulated reflection data were used to create the image in Figure b. In both of these images, we can clearly make out the response due to the glandular structure. To increase the complexity, a 6 mm diameter tumor was added to the BP at x=-8 mm, y= mm and z=-3 mm. The measurement process was repeated, and the image shown in Figure 3 was created. The outline of the glandular insert and tumor are both clearly visible. In future testing, different configurations of the glandular and tumor inserts will be used; however, these preliminary images show how this BP is useful for testing, validation and comparing with simulated data. V. CONCLUSION A new BP for microwave breast imaging has been created. The breast phantom material is made from a carbon/rubber mixture that allows for approximate DPs to be achieved, while maintaining good mechanical properties and stability. The final Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

7 7 Y Position (mm) Y Position (mm) X Position (mm) (a) 4 4 X Position (mm) (b) x x 4 Fig. : Backscatter energy maps created from (a) measured reflection data and (b) simulated reflection data. The BP consists of a skin layer and a 4 cm diameter cylinder. Y Position (mm) X Position (mm) (a) 3 x 4 Fig. 3: Backscatter energy map created from measured reflection data. The cross-section is the z=-3 mm plane. The BP consists of a skin layer, a 4 cm diameter cylinder, and a tumor located at x=-8 mm, y= mm and z=-3 mm BP features a mm skin layer and internal structures that can be easily reconfigured into multiple positions to mimic glandular and tumor tissues. The final BP provides us with realistic skin reflections, and allows for the testing of our imaging algorithm. ACKNOWLEDGMENT This work was supported by Alberta Innovates Health Solutions, Alberta Innovates Technology Futures and Biovantage. The authors would also like to thank Jérémie Bourqui and Zymetrix. REFERENCES [] E. C. Fear, J. Bourqui, C. Curtis, D. Mew, B. Docktor, and C. Romano, Microwave Breast Imaging With a Monostatic Radar-Based System: A Study of Application to Patients, IEEE Trans. Microwave Theory Techn., vol. 6, no. 5, pp. 9 8, May 3. [] J. D. Shea, P. Kosmas, S. C. Hagness, and B. D. Van Veen, Three-dimensional microwave imaging of realistic numerical breast phantoms via a multiple-frequency inverse scattering technique, Med. Phys., vol. 37, no. 8, p. 4 6, Aug.. [3] X. Li, E. Bond, B. Van Veen, and S. Hagness, An overview of ultra-wideband microwave imaging via space-time beamforming for early-stage breast-cancer detection, IEEE Antennas Propag. Mag., vol. 47, no., pp. 9 34, Feb. 5. [4] M. Klemm, J. a. Leendertz, D. Gibbins, I. J. Craddock, A. Preece, and R. Benjamin, Microwave Radar-Based Differential Breast Cancer Imaging: Imaging in Homogeneous Breast Phantoms and Low Contrast Scenarios, IEEE Trans. Antennas Propagat., vol. 58, no. 7, pp , Jul.. [5] D. Li, P. M. Meaney, T. Raynolds, S. a. Pendergrass, M. W. Fanning, and K. D. Paulsen, Parallel-detection microwave spectroscopy system for breast imaging, Rev. Sci. Instrum., vol. 75, no. 7, p. 35 3, Jul. 4. [6] V. Zhurbenko, T. Rubæ k, V. Krozer, and P. Meincke, Design and realisation of a microwave three-dimensional imaging system with application to breast-cancer detection, IET Microw. Antennas Propag., vol. 4, no., p., Dec.. [7] M. Klemm, I. J. Craddock, J. a. Leendertz, A. Preece, and R. Benjamin, Radar-Based Breast Cancer Detection Using a Hemispherical Antenna Array Experimental Results, IEEE Trans. Antennas Propagat., vol. 57, no. 6, pp , Jun. 9. [8] M. Lazebnik, D. Popovic, L. McCartney, C. B. Watkins, M. J. Lindstrom, J. Harter, S. Sewall, T. Ogilvie, A. Magliocco, T. M. Breslin, W. Temple, D. Mew, J. H. Booske, M. Okoniewski, and S. C. Hagness, A large-scale study of the ultrawideband microwave dielectric properties of normal, benign and malignant breast tissues obtained from cancer surgeries, Phys. Med. Biol., vol. 5, no., pp , Oct. 7. [9] M. Lazebnik, L. McCartney, D. Popovic, C. B. Watkins, M. J. Lindstrom, J. Harter, S. Sewall, A. Magliocco, J. H. Booske, M. Okoniewski, and S. C. Hagness, A large-scale study of the ultrawideband microwave dielectric properties of normal breast tissue obtained from reduction surgeries, Phys. Med. Biol., vol. 5, no., pp , May 7. [] A. Guy, Analyses of Electromagnetic Fields Induced in Biological Tissues by Thermographic Studies on Equivalent Phantom Models, IEEE Trans. Microwave Theory Tech., vol. MTT-9, no., pp. 5 4, Feb. 97. [] M. Lazebnik, E. L. Madsen, G. R. Frank, and S. C. Hagness, Tissuemimicking phantom materials for narrowband and ultrawideband microwave applications, Phys. Med. Biol., vol. 5, no. 8, pp , Sep. 5. [] J. C. Y. Lai, C. B. Soh, E. Gunawan, and K. S. Low, Homogeneous and Heterogeneous Breast Phantoms for Ultra-Wideband Microwave Imaging Applications, PIER, vol., pp ,. [3] M. Helbig, K. Dahlke, I. Hilger, M. Kmec, and J. Sachs, UWB microwave imaging of heterogeneous breast phantoms, Biomedical Engineering/Biomedizinische Technik., vol. 57, no. SI- Track-B, p , Aug.. Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.

8 8 [4] C. Hahn and S. Noghanian, Heterogeneous breast phantom development for microwave imaging using regression models, International Journal of Biomedical Imaging, vol., Article ID 8367, pp., Jan.. [5] D. W. Winters, J. D. Shea, E. L. Madsen, G. R. Frank, B. D. Van Veen, and S. C. Hagness, Estimating the breast surface using UWB microwave monostatic backscatter measurements, IEEE Trans. Biomed. Eng., vol. 55, no., pp , Jan. 8. [6] M. J. Burfeindt, T. J. Colgan, R. O. Mays, J. D. Shea, N. Behdad, B. D. Van Veen, and S. C. Hagness, MRI-Derived 3D-Printed Breast Phantom for Microwave Breast Imaging Validation, Antennas Wirel. Propag. Lett., vol., pp. 6 63, Dec.. [7] A. Mashal, F. Gao, and S. C. Hagness, Heterogeneous Anthropomorphic Phantoms with Realistic Dielectric Properties for Microwave Breast Imaging Experiments, Microw. Opt. Technol. Lett., vol. 53, no. 8, pp , Aug.. [8] E. Porter, A. Santorelli, M. Coates, and M. Popovic, Time-domain microwave breast cancer detection: extensive system testing with phantoms, Technol. Cancer Res. Treat., vol., no., pp. 3 43, Apr. 3. [9] J. Croteau, J. Sill, T. Williams, and E. Fear, Phantoms for testing radar-based microwave breast imaging, in ANTEM, Feb. 9, pp. 4. [] M. O Halloran, S. Lohfeld, G. Ruvio, J. Browne, F. Krewer, C. O. Ribeiro, V. C. Inacio Pita, R. C. Conceicao, E. Jones, and M. Glavin, Development of anatomically and dielectrically accurate breast phantoms for microwave imaging applications, in Proc. SPIE 977, Radar Sensor Technology XVIII, vol. 977, Baltimore, Maryland, USA, May 4, pp. 7. [] S. M. Salvador and G. Vecchi, Experimental Tests of Microwave Breast Cancer Detection on Phantoms, IEEE Trans. Antennas Propagat., vol. 57, no. 6, pp. 75 7, Jun. 9. [] A. H. Golnabi, P. M. Meaney, N. R. Epstein, and K. D. Paulsen, Microwave imaging for breast cancer detection: advances in three dimensional image reconstruction, in Conf. Proc. IEEE Eng. Med. Biol. Soc., vol., no., pp , Jan.. [3] N. R. Epstein, A. H. Golnabi, P. M. Meaney, and K. D. Paulsen, Microwave dielectric contrast imaging in a magnetic resonant environment and the effect of using magnetic resonant spatial information in image reconstruction, in Conf. Proc. IEEE Eng. Med. Biol. Soc., vol., pp , Jan.. [4] E. Porter, J. Fakhoury, R. Oprisor, M. Coates, and M. Popovic, Improved Tissue Phantoms for Experimental Validation of Microwave Breast Cancer Detection, in Proc. of the Fourth European Conf. on Antennas and Propagation (EuCAP), Barcelona, Spain, Apr., pp [5] J. Bourqui, J. Garrett, and E. Fear, Measurement and Analysis of Microwave Frequency Signals Transmitted through the Breast, International Journal of Biomedical Imaging, vol., Article ID 56563, pp., Jan.. [6] T. Kobayashi, T. Nojima, K. Yamada, and S. Uebayashi, Dry phantom composed of ceramics and its application to SAR estimation, IEEE Trans. Microw. Theory Technol., vol. 4, no., pp. 36 4, Jan [7] H. Tamura, Y. Ishikawa, T. Kobayashi, and T. Nojima, A dry phantom material composed of ceramic and graphite powder, IEEE Trans. Electromagn. Compat., vol. 39, no., pp. 3 37, May 997. [8] J. Chang, M. Fanning, P. Meaney, and K. Paulsen, A conductive plastic for simulating biological tissue at microwave frequencies, IEEE Trans. Electromagn. Compat., vol. 4, no., pp. 76 8, Feb.. [9] Y. Nikawa, M. Chino, and K. Kikuchi, Soft and dry phantom modeling material using silicone rubber with carbon fiber, IEEE Trans. Microwave Theory Techn., vol. 44, no., pp , 996. [3] C. Gabriel, Tissue equivalent material for hand phantoms, Phys. Med. Biol., vol. 5, no. 4, pp. 45, Jul. 7. [3] J. Garrett and E. Fear, Stable and Flexible Materials to Mimic the Dielectric Properties of Human Soft Tissues, Antennas Wirel. Propag. Lett., vol. 3, pp , Mar. 4. [3] D. Popovic, L. McCartney, C. Beasley, M. Lazebnik, M. Okoniewski, S. Hagness, and J. Booske, Precision open-ended coaxial probes for in vivo and ex vivo dielectric spectroscopy of biological tissues at microwave frequencies, IEEE Trans. Microwave Theory Techn., vol. 53, no. 5, pp. 73, May 5. [33] S. Gabriel, R. W. Lau, and C. Gabriel, The dielectric properties of biological tissues III: Parametric models for the dielectric spectrum of tissues, Phys. Med. Biol., vol. 4, no., pp. 7 93, Nov [34] J. Bourqui, M. Okoniewski, and E. C. Fear, Balanced Antipodal Vivaldi Antenna With Dielectric Director for Near-Field Microwave Imaging, IEEE Trans. Antennas Propagat., vol. 58, no. 7, pp , Jul.. [35] J. Bourqui and E. C. Fear, Systems for Ultra-wideband Microwave Sensing and Imaging of Biological Tissues, in Proc. of the Seventh European Conf. on Antennas and Propagation (EuCAP), Gothenburg, Sweden, 3, pp [36] J. Sill and E. Fear, Tissue Sensing Adaptive Radar for Breast Cancer Detection Experimental Investigation of Simple Tumour Models, IEEE Trans. Microwave Theory Techn., vol. 53, no., pp , Nov. 5. John Garrett (S ) received his BSc degree in electrical engineering from the University of Alberta in, and his MSc degree in electrical engineering from the University of Calgary in 4. He is currently studying towards a DPhil degree in Astrophysics at the University of Oxford where he is developing new technologies for millimeter-wave and THz receivers. His research interests include applied electromagnetics, biomedical imaging and farinfrared detector development. Mr. Garrett s research has been recognized through the Alberta Innovates Technology Futures scholarship, the Clarendon Fund scholarship, and the IEEE Antennas and Propagation Society Pre-Doctoral Research Award. Elise Fear (S 98-M -SM ) is a Professor in the Department of Electrical and Computer Engineering at the University of Calgary and the AITF icore Strategic Chair in Multi-modality Imaging and Sensing. Dr. Fear completed her BASc in Systems Design Engineering at the University of Waterloo in 995, followed by her MAsc and PhD in Electrical Engineering at the University of Victoria in 997 and. She joined the University of Calgary as an NSERC Postdoctoral Fellow in and as a faculty member in. Dr. Fear s research in imaging and sensing includes development of new sensors and enhanced algorithms, as well as prototype system implementation. A key focus is exploring the use of microwaves in medical applications, and her group has performed initial testing of prototype systems on patients. Her contributions have been recognized with the 7 Outstanding Paper Award from the IEEE Transactions on Biomedical Engineering, Top 4 under 4 from Calgary s Avenue Magazine, and a U of C Killam Interdisciplinary Research Prize. Copyright (c) 5 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by ing pubs-permissions@ieee.org.