An Approach to Classify Digital Mammogram Using GLCM Features

Abstract

Breast cancer is the one of the leading cause of death in women. Early and accurate detection of breast cancer can reduce the death rate effectively. Presently, mammography is the most reliable and effective technique in detection of suspicious masses. In digital mammogram, masses are early signs of breast cancer. The computer aided diagnosis (CAD) tool helps the radiologist in detecting the suspicious mass effectively in digital mammogram. But due to the diversity in shape, size, ambiguous margins and poor contrast between the mass and surrounding bright structure, detection and classification of suspicious mass is challenging. This paper presents an approach to classify digitized mammogram from the detected suspicious mass using GLCM features. The GLCM features are extracted after preprocessing, segmentation and region of interest ROI (suspicious mass) detection of the digital mammogram. Preprocessing is preliminary stage used in mammogram image enhancement. It detects and removes the unwanted labels and signs present on the mammogram and extract the breast region. Once the breast region has been found, pectoral muscle and side black strips are detected and eliminated. Then segmentation process is carried out to detect the suspicious mass in digital mammogram. GLCM features are extracted from the detected mass. Analysis of these GLCM features can provide clues and can be used to classify the digital mammogram as benign or malignant.