Microscopic Biopsy Image Segmentation Using Hybrid Color K-Means Approach

Abstract

The color image segmentation is a fundamental requirement for microscopic biopsy image analysis and disease detection. In this paper, a hybrid combination of color k-means and marker control watershed based segmentation approach is proposed to be applied for the segmentation of cell and nuclei of microscopic biopsy images. The proposed approach is tested on breast cancer microscopic data set with ROI segmented ground truth images. Finally, the results obtained from proposed framework are compared with the results of popular segmentation algorithms such as Fuzzy c-means, color k-means, texture based segmentation as well as adaptive thresholding approaches. The experimental analysis shows that the proposed approach is providing better results in terms of accuracy, sensitivity, specificity, FPR (false positive rate), global consistency error (GCE), probability random index (PRI), and variance of information (VOI) as compared to other segmentation approaches.