Mammogram Segmentation Methods: A Brief Review

Being the prime reason, after skin cancer, of high mortality rate among women in present day, breast cancer requires correct diagnosis and precise treatment at its earliest stage. From the time of the advent of diagnosis tools, medical practitioners have left no stone unturned in their efforts of delivering timely medication to the patients; but often human error has resulted in either death due to dosage of medicines resulting from wrongly detected malignancies or due to negligence arising from not detecting the tumors at the right time. Hence, computer-aided diagnosis (CADx) has come into light as a key tool in statistically analyzing medical images obtained from various imaging machines and classifying the specimens into the categories of normal, benign, and malignant. A major step involved in it is the segmentation of the medical image into various regions and determining the required region-of-interest (ROI) from them. Automated image segmentation is quintessential today in order to extract the correct suspicious regions for diagnosis, instead of relying on erroneous human eye judgment. The following study aims to compare and analyze the effectiveness of some existing segmentation methods used to extract the ROIs for analysis of digital mammograms for breast cancer detection.