Comparison of Watershed and Local Center of Mass Techniques in Segmentation of Breast Mammogram

Tumor interpretation on cancer image can lead to early detection of cancer with the prevalence of breast cancer being the second leading cause for death in women. Segmentation of mammographic images is a challenging space because of the complexity in extracting data without causing any source of artefacts on the image. Diagnosis becomes difficult when data extraction is challenging. In order to cater effective diagnosis and effective treatment, segmentation is a vital process. This paper discusses about two segmentation method, Watershed and Local Center of Masses. Comparison between these two algorithms based on the amount of data extraction for six different categories of mammographic images, the apt segmentation method for data extraction is found out. Watershed provides an average dice score of 0.53, and occupies 665 KB memory space and avails 5 seconds running time whereas LCM avails on an average of 0.59 dice score, 250 KB memory space and 1008 seconds computing time. This can...