Detection of microcalcification in mammogram images using semi-automated texture based Grabcut Segmentation

This paper provides a new segmentation technique to detect microcalcifications in mammogram images. It describes the principle of modified GrabCut Segmentation technique. This new GrabCut technique uses k-Means algorithm instead of Gaussian Mixture Model (GMM) to compute the clusters. The texture component is integrated into this GrabCut framework to detect the texture-based boundaries that enables a highly accurate and efficient segmentation.