This paper presents a new artificial neural network (ANN)-based radar data processing approach for the detection of breast cancers located inside fibro-glandular tissues. The aim is not the breast imaging but detecting tumors through ANN processing of data extracted from the radar signals measured around the breast. The proposed approach has been assessed using several realistic two-dimensional breast geometries derived from the models provided by the numerical breast phantom repository of the University of Wisconsin Cross-Disciplinary Electromagnetic Laboratory (UWCEM). The pulsed radar system was assumed to operate in the mono-static configuration. The obtained results showed the abilities of the proposed approach to detect, for any single radar trace, tumors located inside the fibro-glandular tissues with a sensitivity of 93%, a specificity of 90%, and an overall accuracy of 92%.
