Active Contour Based Micro-Calcification Detection and Classification in Digital Mammogram

Abstract

Micro-calcifications are important signs of breast cancer. Detection of calcifications and dense masses becomes a significant problem for computer-aided diagnosis (CAD). The proposed work deals with a novel approach for the extraction of features like micro-calcifications and mass lesions in mammograms, and provides classification for the detected cells. The proposed technique is a four-step procedure: (i) regions of interest specification; (ii) improvement of image by two-dimensional wavelet transformations; (iii) feature extraction, the identification of micro-calcification and mass lesions through level set method; and (iv) classification of disease using a back propagation neural classifier. The effectiveness of the proposed CAD scheme is validated from experimental results.