Segmentation of Irrelevant Regions using Color Thresholding Method: application in Breast Histopathology Images

Abstract

Segmentation of irrelevant regions in the breast histopathology image is essentially performed in preliminary or post processing stages. This study presents a color thresholding method to segment the irrelevant regions, specifically blood cells and hemorrhage in the RGB breast histopathology images. The conventional gray level global thresholding method was extended to color images by modifying the thresholding algorithm. The color thresholding method was performed by decomposing the RGB input image into coherent regions based on similarity in color feature. Area Overlap Measure (AOM) and Combined Equal Importance (CEI) are two statistical metrics used to evaluate the performance of the color thresholding method. The obtained AOM and CEI for the overall dataset in blood cells are 0.92 and 0.89, respectively. The color thresholding method is found to be effective and is able to produce plausible thresholding results in segmentation of irrelevant regions in RGB breast histopathology images.