Sub-Wavelength Microwave Radar Imaging for Detection of Breast Cancer Tumors

Abstract

Ultra-wideband microwave imaging, with its advantages of absence of breast compression, nonionizing and noninvasive properties, is a complementary method to X-ray mammography for breast cancer detection that is safe and reasonably inexpensive. The motivation for employing microwave imaging techniques for detecting early-stage breast cancer stems from published research results showing the strong contrast in the dielectric properties at microwave frequencies between normal breast tissue and malignant lesion. This thesis contributes to development of novel techniques for the detection of early-stage breast cancer tumors well below a centimeter with specificity and high degree of accuracy, i.e., with minimum false negatives/positives. In our proposed approach, a modified Shannon entropy criterion (SEC) is formulated for determining when the time-reversed wave focuses back to the source target in the presence of an inhomogeneous lossy medium. It is demonstrated through two examples, the time-reversal mirror and cavity, that the SEC is found to be more robust than the inverse varimax norm. TR has been shown to be superior to other simple delay-based focusing techniques and here we have extended the TR algorithm by making it more robust in localizing small tumors. The importance of this finding becomes evident as the SEC allows for the detection of tumors that are sub-wavelength in size. Our novel sub-wavelength ultra-wide band (UWB) microwave radar imaging technique exploits the principle of phase-shifting mask (PSM) from optical lithography and is implemented using a time-reversal (TR) algorithm based on the transmission-line matrix (TLM) method. We incorporate the SEC in a TR algorithm to achieve a robust imaging algorithm exploiting the measurements acquired by our phase-shifting mask (PSM) experimental set-up. Unlike the FDTD TR algorithm by Kosmas et al., which excites one of the 23 antenna elements, we propose a different system where all antennas are stimulated simultaneously with their excitation based on the PSM principle. A 0.5-mm diameter tumor was detected and located using a 200-ps UWB pulse in a realistic inhomogeneous two dimensional breast model. The breast model was derived from magnetic resonance imaging data and simulated using the TLM method. The effect of the dielectric contrast and proximity of tumor to the antenna receivers are examined. A TLM-based TR algorithm employing two types of time reversal mirrors (TRMs) is proposed to improve the accuracy of localizing the sub-wavelength tumors. The final part of the thesis examines the feasibility and the design of a narrowbeam UWB antenna for microwave breast cancer detection focusing on the antenna feed structure and the printed TEM horn antenna. A feed structure of an UWB antenna for microwave radar imaging is designed. The UWB antennas are fundamental components of the UWB radar imaging hardware. The performance of the antenna is crucial for the resolution and the reliability of the whole imaging system. A TEM horn antenna is studied and suggestions are made regarding the hardware implementations of the experimental setup. We conclude by suggesting future work toward hardware and practical implementation of UWB microwave radar imaging with high resolution.