Breast cancer detection: Comparison of data-dependent and data-independent approaches

This paper addresses the early detection of malignant breast tissues using microwave imaging. The breast contains fibroconnective, glandular and benign tissues whose dielectric properties can be very close to those of malignant tissues. Thus it becomes difficult to differentiate the malignancy using conventional microwave imaging approaches. Here, we investigate RCB and DAS methods coupled with Matrix Pencil method to detect and differentiate the malignant tissues from other tissues. The RCB and DAS techniques are employed to obtain the images that localize suspected regions. Then Complex Natural Resonance (CNR) from those regions are estimated that correspond to the morphological properties of benign and malignant tissues. The performance of RCB and DAS beamforming approaches are investigated and compared while the Matrix Pencil method is used for CNR extraction. The simulation results are presented using numerical 2-D FDTD breast model.