Gradient Watershed Transform Based Automated Cell Segmentation for THG Microscopy Medical Images to Detect Skin Cancer

Abstract

The conventional cell segmentation by manual selection of cells takes several minutes to segment and analyze one image and may suffer from fatigue of human eyes and subjective judgments by medical staff resulting in poor results. In this paper, a computer-aided design for automatic cell segmentation and NC ratio analysis was proposed. The proposed cell segmentation approach, is based on the watershed-based approach and the concept of convergence index filter for automatic cell segmentation. Image thresholding is the simplest method for high-speed cell segmentation, but it produces good results only for images with high contrast between object and background. Watershed transform another common technique, uses the concept of morphological image processing and considers contextual information to produces table segmentation results. Experimental result shows better result when compare with existing one.