Mammogram Analysis Based on Pixel Intensity Mean Features

Abstract

Problem statement: In the recent years, Computer Aided Diagnosis (CAD) can be very useful for detection of breast cancer. Mammography can be used as an efficient tool for breast cancer diagnosis. A computer based diagnosis and classification system can reduce unnecessary biopsy. Approach: This study investigates a new approach to the classification of mammogram images based on pixel intensity mean features. The proposed method for the classification of normal and abnormal (cancerous) pattern is a two step process. The first step is feature extraction. The intensity based features are extracted from the digital mammograms. The second step is the classification process, differentiating between normal and abnormal pattern. Artificial neural networks are used to classify the data. Experimental evaluation is performed on the Digital Database for Screening Mammography (DDSM), benchmark database. Results and Conclusion: Experiments are performed to verify that the proposed pixel intensity mean features improve the accuracy of the classification. The proposed CAD system achieves better classification performance with the accuracy of 98%.