Detection of the cancerous tissue sections in the breast optical biopsy dataflow using neural networks

Abstract

The method of artificial neural networks was applied for analysis of the data obtained in the clinical trials of the optical biopsy system. Detection of malignant tissue sections was carried out using a multilayer perceptron. The coefficients of wavelet decomposition of optical scattering spectra were given at the perceptron input and its output gave the malignancy probability for the current spectrum. End-toend probability calculation throughout the optical biopsy procedure dataset showed reliable detection of the cancer sections in the same place as it was specified by experts.