Breast Cancer Detection using UWB

This article proposed a novel method for detecting breast cancer using Ultra-Wide Band (UWB) antenna used as a sensor capable of capturing complex and small variations in the dielectric properties of the tissue layers of the prepared breast phantom. The antenna, made of RT Duroid 5880 substrate, gathers S11 and S21 signals from the data collection. Discrete Wavelet transformation is performed on the samples and wavelet-based threshold denoising of the approximation and detail coefficients of the transform. The reconstruction of the coefficients is performed using Inverse Discrete Wavelet Transform. A deep-learning model is used to extract deep features from 2D time frequency scalogram images generated by a Cosine Wavelet Transform of the denoised signals. Two benchmark models are used for comparison with the proposed neural model. The accuracies obtained in homogeneous and heterogeneous phantoms are 97.55% and 98.76% for augmented data and 96.78% and 97.45% for un-augmented data.