Improvement in image segmentation analysis in digital mammography

Abstract

Digital Mammography is an effective tool in identifying the tumor cells in breast. Image segmentation plays a major role in identifying the location of benign and malignant tumor cells. Improvement in the area of Image segmentation analysis of Digital mammography leads to improved diagnosis for the presence of tumor cells. Worldwide, Breast cancer is the most common cancer among the female adults. Worldwide Health Organization (WHO), in its report, had mentioned that Cancer is a major cause of death around the world. Breast leads the top sites of cancer occurrences among women. As per the Cancer Research UK, Breast Cancer survival statistics 2015, more than 90% of women diagnosed with breast cancer at the earliest stage survive their disease for atleast five years compared to 15% of women diagnosed with the most advanced stage of disease. Finding and treating cancer at an early stage can save lives. The main purpose of screening healthy women for breast cancer is to diagnose the disease earlier and by doing so, reduce the risk of or delay the onset of death from the disease. Statistically, the stage at which breast cancer is identified has an impact on the Complete Response for its treatment. Early detection of the breast cancer leads to proper treatment of cancer at early stage itself and ensures good survival rate. Expertise and competence of Radiologists are the deciding factors in effective digital mammogram analysis and findings. With advancements in image processing and machine learning, automated detection of microcalcification clusters from mammogram images are proposed in thiswork. The objective of the thesis is to propose a segmentation algorithm by which mammogram is labeled as benign, malignant or normal breast. Challenges and Issues inherent in analyzing the breast image for microcalcifications include: (a) Intensities from background of breast image interferes in classifying glandular and fatty region within the breast tissue, (b)Presence of artifacts and templates in scanned mammogram image produce