Computer Aided Diagnosis System for Classification of Mammogram Using Deep Learning Algorithm

Develop a completely automated CAD (computer aided diagnosis) system for full mammogram screening for breast cancer diagnosis is still challenging problem. In most of the cases, full mammogram based CAD system comprises of following processing segments; pre-processing of raw images, image enhancement, ROI detection, two stage mass classification, at first stage normal or abnormal mass classification and at second stage abnormal mass is further identified as benign or malignant. For BI-RADS classification, Automated CAD systems have not obtained the required efficiency. An important factor for detecting the mass and classifying it, is breast density. The core concept behind this work is to achieve a desired accuracy for classifying mass using proposed work of completely automated system more intended on breast mass density. Texture quantification of the full mammogram images is used for measuring breast mass density. Proposed completely automated CAD system framework based on mainly Breast mass density(calculated by texture quantification) using the mammographic image dataset and also explored the other approaches in the same direction.