Novel technique for mammogram image segmentation using moment preserving

Breast cancer detection is one of the crucial research problems in which mammogram image segmentation plays key role. Thresholding is one of the widely used segmentation approach. In this approach the challenge is to calculate the optimal threshold that separates the lesion from the background (mainly tissue). Various thresholding techniques have been proposed. Using moment preserving principle an image can be segmented into meaningful gray level segments while preserving moments of the original image. On the other hand Gamma function better models the distribution of gray levels in a mammogram. Exploiting moment preserving principle and Gamma distribution, a novel thresholding technique for the segmentation of mammograms is proposed. We tested our algorithm on several DDSM database images and found that it segments the lesion part from the background in a better way.