NUMERICAL AND PHYSICAL MODELS FOR MICROWAVE BREAST IMAGING

Abstract

<p>Breast cancer is a serious health problem that severely affects the society. Science and engineering communities are looking to find more advanced and new cures to contain this deadly disease. There has been considerable success in increasing the efficacy of conventional imaging techniques for breast cancer detection such as mammography. However, the use of ionizing radiation, the incidence of patient discomfort and the low specificity are some of the negative aspects of mammography. Also, the chances to successfully cure breast cancer can increase manifold with early detection of the malignant tumors. Magnetic Resonance Imaging (MRI) is successful with its highly clear and detailed images, but it is considered too expensive for wide screening. Its low specificity is another disadvantage.</p> <p>Microwave imaging of the breast, which uses non-ionizing radiation, is a<br />very active research area of recent times. It is hoped that once developed, it can override the negative aspects of the existing imaging modalities. The idea is to detect malignant lesions in the breast based on their electrical-property contrast with healthy breast tissues.</p> <p>The electrical properties of biological tissues in the microwave region are described by their complex permittivity. Previous studies claimed that malignant tumors have very high permittivity and high loss compared to the surrounding healthy tissue. It was assumed that the modality exploiting this contrast can easily detect sub-centimeter tumors. Hence, very simple models with high permittivity and loss tumor models embedded in low permittivity homogenous models were employed to show the detection capabilities of microwave imaging techniques. The examples were fairly amateurish and claimed unrealistic performance of the rather simplistic microwave imaging techniques.</p> <p>Recent studies have shown that the breast is not as simple an electromagnetic medium as was thought previously. It is a complex heterogeneous environment with high permittivity and high loss healthy tissues (such as glandular tissue) and low permittivity and low loss healthy tissues (such as fat). The internal tissue distribution and volume varies significantly from woman to woman. Thus, simple models made with hemispheres, cylinders, etc., filled with water or oil cannot mimic the reality of the breast in imaging experiments and simulations.</p> <p>This is why there is a considerable demand of representing breasts as realistic physical and numerical models. Physical models are required to experimentally verify the imaging scanning systems with antennas. The numerical models are required for electromagnetic (EM) simulations that are indispensable in designing and testing image reconstruction algorithms.</p> <p><br />The work in this thesis mainly deals with the construction and representation of breast models both in experiments and in simulations. This is impoliant work toward making microwave imaging a practical stand-alone detection system.</p>