Identification of Abnormal Masses in Digital Mammogram Using Statistical Decision Making

The increasing threat of breast cancer in developing countries may not only be handled by the existing medical setup as well as insufficient number of medical workforces. To handle the increasing volume of data produced by diagnostic imaging that can be efficiently managed by computer aided detection/diagnosis (CAD) to assist medical practitioners in image interpretation to detect structural abnormalities like tumour. Mammography has been proven to be the most reliable and cost-effective methodology for early breast tumor detection. In this paper, an abnormality detection methodology has been proposed alongwith preparation and pre-processing steps. The accuracy of CAD to detect abnormalities on medical image analysis depends on a robust segmentation algorithm. Here two types of segmentation mechanism have been implemented i.e. edge-based and region-based. Finally, a proposed statistical decision-making system is used to extract the abnormal region(s) based on intensity distribution. Applying the proposed method on CR and DR mammographic images produces the quantitative measures accuracy, sensitivity and specificity as 96%, 97.6% and 88.6% respectively which is comparable with other contemporary research works.