Anastasia Baran

Breast Imaging Using Microwave Tomography with Radar-Based Tissue-Regions Estimation

Microwave tomography (MWT) and a radar-based region estimation technique are combined to create a novel algorithm for biomedical imaging with a focus on breast cancer detection and monitoring. The region estimation approach is used to generate a patient-specific spatial map of the breast anatomy that includes skin, adipose and fibroglandular regions, as well as their average dielectric properties. This map is incorporated as a numerical inhomogeneous background into an MWT algorithm based on the finite element contrast source inversion (FEM-CSI) method. The combined approach reconstructs finer structural details of the breast and better estimates the dielectric properties than either technique used separately. Numerical results obtained with the novel combined algorithmic approach, based on synthetically generated breast phantoms, show significant improvement in image quality.

Integrating prior information into microwave tomography Part 1: Impact of detail on image quality

Purpose: The authors investigate the impact that incremental increases in the level of detail of patient‐specific prior information have on image quality and the convergence behavior of an inversion algorithm in the context of near‐field microwave breast imaging. A methodology is presented that uses image quality measures to characterize the ability of the algorithm to reconstruct both internal structures and lesions embedded in fibroglandular tissue. The approach permits key aspects that impact the quality of reconstruction of these structures to be identified and quantified. This provides insight into opportunities to improve image reconstruction performance. Methods: Patient‐specific information is acquired using radar‐based methods that form a regional map of the breast. This map is then incorporated into a microwave tomography algorithm. Previous investigations have demonstrated the effectiveness of this approach to improve image quality when applied to data generated with two‐dimensional (2D) numerical models. The present study extends this work by generating prior information that is customized to vary the degree of structural detail to facilitate the investigation of the role of prior information in image formation. Numerical 2D breast models constructed from magnetic resonance (MR) scans, and reconstructions formed with a three‐dimensional (3D) numerical breast model are used to assess if trends observed for the 2D results can be extended to 3D scenarios. Results: For the blind reconstruction scenario (i.e., no prior information), the breast surface is not accurately identified and internal structures are not clearly resolved. A substantial improvement in image quality is achieved by incorporating the skin surface map and constraining the imaging domain to the breast. Internal features within the breast appear in the reconstructed image. However, it is challenging to discriminate between adipose and glandular regions and there are inaccuracies in both the structural properties of the glandular region and the dielectric properties reconstructed within this structure. Using a regional map with a skin layer only marginally improves this situation. Increasing the structural detail in the prior information to include internal features leads to reconstructions for which the interface that delineates the fat and gland regions can be inferred. Different features within the glandular region corresponding to tissues with varying relative permittivity values, such as a lesion embedded within glandular structure, emerge in the reconstructed images. Conclusion: Including knowledge of the breast surface and skin layer leads to a substantial improvement in image quality compared to the blind case, but the images have limited diagnostic utility for applications such as tumor response tracking. The diagnostic utility of the reconstruction technique is improved considerably when patient‐specific structural information is used. This qualitative observation is supported quantitatively with image metrics.

A faceted magnetic field probe resonant chamber for 3D breast MWI: A synthetic study

The development of a 3D Microwave Imaging system for breast cancer detection is described. The system is comprised of twenty four magnetic field probes mounted in an air-filled faceted resonant chamber. Simulations of the system are detailed, and the ability to image a simplified four region breast phantom using appropriate prior information is demonstrated. Inversion of the data was performed by a full vectorial 3D FEM-CSI algorithm. Prototype magnetic field probes are described and their performance is evaluated using GTEM cell measurements. The simulation and probe performance results lay the foundation for continued development of this system.

Immersion medium independent algorithm for breast microwave imaging

Biomedical imaging at microwave frequencies has shown potential for breast cancer detection and monitoring. The Universities of Manitoba and Calgary have recently developed an algorithm that combines microwave tomography (MWT) and radar imaging by incorporating radar-derived regional permittivity maps as prior information into a finite element contrast source inversion (FEM-CSI) algorithm (Baran et. al., Pier, 149, 161–171, 2014). Matching the immersion medium to the dielectric properties of the breast is often regarded as crucial for adequate image reconstruction, however we demonstrate that high quality images are possible in any immersion media when the combined radar/MWT approach is used.

Breast cancer imaging using microwave tomography with radar-derived prior information

Summary form only given. Biomedical imaging at microwave frequencies has shown potential for breast cancer detection and monitoring. Current modalities suffer from significant underlying disadvantages. For example, mammography utilizes high-energy, ionizing radiation and is uncomfortable for patients, and breast MRI has a high false positive rate due to high sensitivity and low specificity. Microwave imaging is an inexpensive technique that uses low-power, non-ionizing radiation that is not harmful to patients. Two techniques that exploit microwave frequencies for breast imaging are microwave tomography (MWT) and radar-based imaging. These two techniques suffer from limitations in resolution of fine structures and accuracy of tissue dielectric properties.We present a novel algorithm that combines MWT with a radar-based region estimation technique, with a focus on breast cancer imaging. The region estimation method creates a patient-specific spatial map of the breast anatomy that includes skin, adipose and fibroglandular tissue regions, and contains the average dielectric properties over those regions (D. Kurrant and E. Fear, Inverse Prob., 2012). This map is incorporated into a finite element contrast source inversion (FEM-CSI) algorithm as prior information in the form of an inhomogeneous background (A. Zakaria, A. Baran, and J. LoVetri, Antennas Wireless Propag. Lett., 2012). This hybrid approach is able to reconstruct finer structural details of tissues within the breast, and estimates their dielectric properties more accurately than either technique used alone. Results from various numerical phantoms characterize this significant improvement in image quality. In addition to improvement of accuracy and resolution, the algorithm is able to produce reliable results within the 1GHz-4GHz frequency range, allowing us to take advantage of march-on-frequency techniques to further improve image quality and localize tumors. Simulations also produce reliable results using several different immersion media that vary greatly in their real and imaginary permittivity values, providing more flexibility in the choice of immersion medium for clinical systems. Results from numerical breast phantoms using this hybrid technique will be presented and compared to traditional MWT methods.

Integrating prior information into microwave tomography part 2: Impact of errors in prior information on microwave tomography image quality

Purpose: The authors have developed a method to combine a patient‐specific map of tissue structure and average dielectric properties with microwave tomography. The patient‐specific map is acquired with radar‐based techniques and serves as prior information for microwave tomography. The impact that the degree of structural detail included in this prior information has on image quality was reported in a previous investigation. The aim of the present study is to extend this previous work by identifying and quantifying the impact that errors in the prior information have on image quality, including the reconstruction of internal structures and lesions embedded in fibroglandular tissue. This study also extends the work of others reported in literature by emulating a clinical setting with a set of experiments that incorporate heterogeneity into both the breast interior and glandular region, as well as prior information related to both fat and glandular structures. Methods: Patient‐specific structural information is acquired using radar‐based methods that form a regional map of the breast. Errors are introduced to create a discrepancy in the geometry and electrical properties between the regional map and the model used to generate the data. This permits the impact that errors in the prior information have on image quality to be evaluated. Image quality is quantitatively assessed by measuring the ability of the algorithm to reconstruct both internal structures and lesions embedded in fibroglandular tissue. The study is conducted using both 2D and 3D numerical breast models constructed from MRI scans. Results: The reconstruction results demonstrate robustness of the method relative to errors in the dielectric properties of the background regional map, and to misalignment errors. These errors do not significantly influence the reconstruction accuracy of the underlying structures, or the ability of the algorithm to reconstruct malignant tissue. Although misalignment errors do not significantly impact the quality of the reconstructed fat and glandular structures for the 3D scenarios, the dielectric properties are reconstructed less accurately within the glandular structure for these cases relative to the 2D cases. However, general agreement between the 2D and 3D results was found. Conclusion: A key contribution of this paper is the detailed analysis of the impact of prior information errors on the reconstruction accuracy and ability to detect tumors. The results support the utility of acquiring patient‐specific information with radar‐based techniques and incorporating this information into MWT. The method is robust to errors in the dielectric properties of the background regional map, and to misalignment errors. Completion of this analysis is an important step toward developing the method into a practical diagnostic tool.

Effects of data collection schemes and systems on the imaging performance of electromagnetic inverse problems

Summary form only given. Although electromagnetic inverse problems are, as is canonically understood, ill-posed mathematical problems, there are several possibilities that arise with respect to the practical design of the imaging system with which scattered field data is collected that can ameliorate the problem and significantly improve the imaging performance. The standard techniques for dealing with the ill-posedness of the problem aim at reducing modelling error and regularizing the mathematical inverse problem. This includes the creation of simplified systems that are amenable to a manageable numerical inversion model; both in terms of the achievable model accuracy as well as with respect to the required computational resources. Of course, the reduction of modelling errors is important as large errors between the numerical model and the actual system contribute to the inability of the inversion algorithm to converge to the true solution (i.e., modelling error manifests itself as systematic non-random noise and the instability of the inverse problem results in convergence to non-unique, non-true, solutions of the inverse problem). Data calibration techniques and numerical methods of regularizing the mathematical problem have been well studied in the past, and are indeed necessary to arrive at useful solutions. In this work we focus on some available system design options that can promote better convergence and accuracy of the converged solution. In particular, the following options will be considered: 1) the use of resonant metallic chambers of various shapes; 2) the collection of different field components within the chamber; 3) the use of several immersion media; and 4) the use of dynamic bound-aries to establish not only diverse incident field data, but also to diversify the effective Greens function of the inverse problem. With regard to the first option, the way one incorporates the boundary conditions of the chamber into the inversion algorithm will be delineated. It will be shown that several options are available with respect to the formulation of the data equation and regularization terms. Under the second option, we will show how collecting the tangential magnetic field on the surface of the chamber walls can provide advantages with respect to modelling error and the amount of data that can be collected. Under the third option, we show how the inversion algorithm can be made quasi-independent of the immersion medium that is used (when, for example, in breast imaging, such a design parameter is available) and we will show how for the same frequency the use of different immersion media will provide a variety of diverse data that interrogate the object of interest with different wavelengths. Finally, with the use of dynamic boundaries that can be turned off and on, we show how a different incident field can be produced for the exact same transmitter antenna location. For the most part we focus on biomedical imaging applications where all of these options are available, as well as on a novel stored-grain imaging application that is implemented within grain bins.

Iterative refinement of fibroglandular region with microwave breast imaging

A procedure is proposed in the context of microwave breast imaging to iteratively refine the resolution of the reconstruction model for the glandular region used by the imaging algorithm. The reconstruction model begins with a small number of large reconstruction elements and the dielectric properties are estimated over these elements. For subsequent iterations, the density of the mesh elements within the glandular region increases. The imaging algorithm uses the results from the reconstructed image of the previous step to initialize the dielectric properties of the model elements. An incident field with a higher frequency than the previous step is used. Dielectric properties over the finer model elements are estimated from the scattered data which have a higher spectral content. The procedure continues until a satisfactory resolution of the reconstructed image is achieved. The aim of the strategy is to improve the resolution of the reconstructed profile while overcoming problems associated with solving a highly non-linear and severely ill-posed inverse scattering problem. The feasibility of the algorithm is evaluated with numerical results based on a synthetically generated breast phantom.

A study of contrast-enhanced functional microwave imaging

Microwave imaging (MWI) continues to develop as a low-cost portable complementary soft-tissue imaging modality, particularly in the context of breast cancer detection. Despite dramatic advancements in algorithmic development and signal acquisition, even the most recent imaging studies have shown that challenges still remain to improve this emerging technologys spatial and contrast resolution for anatomical imaging. However, MWIs ability to detect unique properties dependent on the physiological state of a tissue of interest at non-ionizing frequency ranges suggest it may be suitable for safer, cheaper functional imaging studies using non-radioactive contrast agents. This study explores applications in niches traditionally filled by nuclear medicine, where contrast-enhanced MWI could achieve resolutions comparable to existing imaging procedures but with no associated radiation dose. Non-toxic compounds that exhibit strong microwave-band responses, notably transition metal nanoparticles (S. Semenov et al., IFMBE Proc. 25/8, 311-313, 2009) and free radicals, may have promise in such contrast-enhanced imaging to provide metabolic rather than strictly anatomic data. To fully exploit the information available from these agents, the addition of an external weak polarizing magnetic field (PMF) across the imaging domain is necessary, which primarily influences ferromagnetic or strongly paramagnetic contrast agents within that domain (O.M. Bucci et al., IEEE Trans. Biomed. Eng., 58, 9, 2528-2536, 2011). Along with the traditionally measured changes in permittivity and conductivity, the PMF allows variations in magnetic susceptibility to contribute to the relevant microwave image data through resonance phenomena (P.C. Fannin, J. Mol. Liquids, 114, 79-87, 2004).

Simultaneous high-order contrast source inversion of dielectric and magnetic targets

Magnetic contrast agents have been recently proposed as a method of improving the capabilities of microwave imaging for cancer diagnosis, detection and treatment monitoring. In order to exploit these contrast agents, electromagnetic inversion algorithms should be based on forward solvers capable of predicting the scattered fields from both dielectric and magnetic targets. To this end we have developed a high-order, nonlinear inversion algorithm for the simultaneous inversion of magnetic and dielectric targets using the contrast source inversion (CSI) formulation of the inverse problem. The inverse solver uses a high-order, time-harmonic, discontinuous Galerkin formulation of Maxwells equations and supports unstructured discretizations of dielectric, magnetic and perfectly conducting media. The resulting CSI formulation is an unstructured, high-order extension of an existing dielectric and magnetic CSI formulation (A. Abubakar and P. M. van den Berg, J. Comput. Phys., 195(1), 236-262, 2004), and extends FEM-CSI (A. Zakaria, C. Gilmore and J. LoVetri, Inverse Probl., 26(11), 115010, 2010) to both high-order and magnetic materials. In this work we will focus on the modifications to the CSI formulation required to support independent expansion orders for the contrast, contrast sources and fields. High-order contrast expansions effectively decouple the solution from the underlying discretization and, for the same level of accuracy, reduce the number of degrees of freedom in the iterative inversion process. An exact radiating boundary condition has been implemented for open problems and, at the cost of computational time and memory, yields an error-controllable forward solver for electromagnetic inversion. The reconstructions of both dielectric and/or magnetic targets will be presented for two-dimensional image reconstruction of synthetic and experimental data.