Combined segmentation technique for suspicious mass detection in Mammography

Abstract

Breast cancer is one of the prevalent causes of death among women worldwide. Early detection may save life of women and helps for timely recovery. There are several screening tests are available to detect the cancer such as Ultrasound, Mammography and MRI. Among these, Mammography is the most effective screening test for breast cancer. The purpose of screening was to detect abnormalities at an early stage, which may require monitoring and treatment. Mass and microcalcifications are basic signs of abnormalities found using screening test. However it is very difficult to find the mass and its spread in the mammogram. Therefore Computer Aided Diagnosis (CAD) helps radiologist to analysis the abnormality in the mammography without much difficulty. The proposed method aims to segment the suspicious mass from the mammography using combined technique of using watershed, morphological operations and active contour based segmentation techniques. Here we show that, proposed segmentation technique helps to minimize the over segmentation of conventional watershed segmentation technique. The efficiency of the algorithm is measured with Mini-MIAS database. The results reported satisfactory segmentation using proposed contour based segmentation technique. The accuracy of the algorithm is reported to be 95.86% in identification of mass in mammography.