Study On Early Breast Cancer Detection Via UWB Microwave Technology

Breast cancer has become one of the primary cancers which are harmful to thewomen health. The early diagnosis of breast cancer is essential for the long-termsurvival of the breast cancer patients. X-ray mammography, ultrasonic testing,magnetic resonance imaging and CT scanning are treated as the common method forbreast cancer detection. Although there are some advantages of these methods, somelimitations such as the high false detecting rate, radiation to the human body andhigher cost to constraint their application as a regular screening for early breast cancerdetection. Recently, the ultra-wideband (UWB) microwave near-field imagingtechnology is developed for the early breast cancer detection which is based on thehigh contrast in the electric properties of malignant tumor related to the normal fattybreast tissues at microwave frequencies. This technology has obvious advantages suchas accurate, safe, inexpensive, high imaging resolution and adequate penetratingdepth.This project is supported by the National Natural Science Foundation of China(No.61271323) and the open project from State Key Laboratory of Millimeter Waves,China (No.K200913) for the research of early breast cancer detection by UWB.In this paper, the three-dimensional breast model is demonstrated and the finitedifference time domain method (FDTD) is applied to simulate the microwavepropagation in the breast tissues. The dispersive properties of the breast fatty, the skin,the glands are considered for the impact of electromagnetic wave propagation.3D confocal imaging algorithm is applied for the breast cancer detection. In orderto improve the robustness of the algorithm, the norm constrained capon beamforming(NCCB) algorithm and the double constrained robust capon beamforming (DCRCB)algorithm are employed for the imaging reconstruction. The time-shift-delay methodis used for the tumor backscattered signal to improve the imaging sharpness. Thesuperiority of the imaging algorithm is manifested by simulation examples.The late-time component of tumor backscattered signals is analyzed and thematrix pencil method is employed for the pole feature extraction of the tumor withdifferent dielectric properties and different size. The characteristic atom library isestablished by the extracted pole feature of the tumor and the matching pursuit algorithm is applied for the tumor backscatter wave reconstruction based on thecharacteristic atom library. The maximum likelihood estimation method is used forthe identification of the different types of breast tumors. The numerical simulationresults verify the correctness of the method which can be treated as the aideddetection to improve the detection rate of breast tumors.The multiple measurements method with the rotation of the antenna array andthe parallel displacement measurements method by the linear antenna array areapplied for the extraction of reference signals. Simulation results verify thecorrectness of the methods.