Complex Permittivity Reconstruction Using Skin Surface Reflection and Neural Network for Microwave Breast Imaging

Abstract

This study introduces an experimental validation for the complex permittivity profile reconstruction using the multi-layer perceptron (MLP) neural network (NN) approach for quantitative microwave recognition of breast cancer. A direct conversion from the four-dimensional scattered data to the complex permittivity three-dimensional profile can be achieved by combining the MLP-NN and the skin surface rejection preprocessing. The experimental data, measured by ultra-wideband radar equipment using a simplified breast phantom, validates that our approach provides both the real and imaginary parts of complex permittivity profiles, even when using limited numbers of training datasets.