Segmentation of images is one amongst the primitive and most vital stages of processing in images and plays a very crucial role in analyzing medical mammogram images but images of mammogram have the moderate level of distinction and are disrupted with sturdy speckle noise. Owing to the effects, mammogram images segmentation is extremely difficult and conventional segmentation techniques may not lead to result satisfaction. Due to high noise, low distinction, and alternative imaging artifacts, region boundaries in mammogram images often do not adjust to the assumptions of many image processing algorithms. This paper addresses the potencies and weaknesses of the existing techniques of carcinoma detection in mammograms. The paper provides new aspects of research for researchers.