MLO Mammogram Pectoral Muscle Masking with Adaptive MSER

Abstract

Abstract: Breast cancer claims the lives of many people every year. Breast cancer diagnosis is a difficult process that requires competent radiologists. Manual detection of breast cancer disease takes a large amount of time, as does manual treatment of disease. As a result, automated detection is required, which aids in early treatment and in some circumstances, saves lives. These technical advancements are beneficial for early treatment due to resource availability and computing capability. A mammogram is a technique for detecting breast cancer masses early. To identify masses from MLO mammograms, several image processing-based computer-aided diagnostic techniques have been developed. In computer aided diagnosis systems, the presence of Pectoral Muscle in MLO Mammograms has a considerable detrimental influence on mass detection from MLO Mammograms. Masking the Pectoral muscle improves mass detection from an MLO mammogram. Locating the Pectoral muscle is tough since the intensity of this tissue is equivalent to that of a malignancy. The primary goal of this research is to develop a Most Stable Extremal Region (MSER) based method to locate and mask Pectoral muscle from the Mediolateral Oblique Mammogram. The empirical analysis suggests that the proposed novel procedure is straightforward and gives promising results in locating and masking Pectoral muscle. The suggested technique enhances accuracy by 96.27% compared to 95% for state-of-the-art solutions. Python and MATLAB are used to create the new system.

Index Terms: MLO Mammogram, Pectoral muscle, Most Stable Extremal Region (MSER), Image processing