Experimental Study of Breast Cancer Detection Using UWB Imaging

Abstract

Breast cancer detection using UWB imaging is presented in this paper. The study is performed experimentally. Homogeneous breast phantom is constructed using pure petroleum jelly. The tumor is modeled using mixture of water and wheat flour. The breast fatty tissue and tumor tissue are put in breast shaped glass which mimics the skin. The dielectric properties values are comparable to previous study. Neural Network (NN) was trained and tested using feature vector which is prepared by performing discrete cosine transform (DCT) of the received UWB signals. Very encouraging results were obtained. Up to 100 % tumor existence detection was achieved. Tumor size and location detection rate were 91.3% and 95.6% respectively.