Adaptive Gaussian Filtering (AGF) and Segmentation using Deep Neural Network (DNN) for Breast Tumour Classification

Abstract

Breast cancer is one of the major causes of death among women. An improvement of early diagnostic techniques is critical for women‟s quality of life. Mammography is the main test used for screening and early diagnosis. Many researchers worked in the area of breast cancer detection and proposed segmentation methods. However, no solution given by researchers is best promising and has limitations and it is still a challenging problem because of noises presented in the samples. So the early diagnosis of breast cancer plays critical roles in increasing the chance of tumour segmentation results. A few recent works also investigated the digital implementations of Cellular Neural Network (CeNN), they are still computationally redundant and sensitive to input image variations without fully utilizing the advantages of hardware acceleration. This research addresses the problem of segmenting the breast cancer from the Mammography images. Proposed an Adaptive Gaussian Filtering (AGF) algorithm in which the filter variance is adapted to both the noise characteristics and the local variance of the signal. Using an AGF for noise suppression, the noise is smoothed out; at the same time the signal is also distorted. The use of an AGF is proposed as pre-processing for noise removal. Deep Neural Network (DNN) based approach is presented for segmentation. The approach is featured with quantization to significantly reduce the computational complexity and non-linear template for robustness. Compared with other prior works, the proposed work is able to reduce error value and increase segmentation results.