LAUSR

The timely diagnosis of skin lesion diseases is highly difficult for people living in rural or far flung areas due to dearth of qualified dermatologists. However, the dermatologists can diagnose skin lesion diseases by carefully examining the high-quality images at their clinics or from a distance area. Further, the computerized automatic diagnostic system may assist primary health professionals for quick and accurate diagnosis of these skin diseases. Thus, there is a need for medical image processing and analysis of skin lesion images to enhance their visibility properties. An efficient and effective skin lesion detection and identification software tool will provide a better classification system and may enhance the automation of skin lesion diagnosis. In this work, detection of skin lesions from human skin images is conducted by utilizing three image processing segmentation methodologies namely—Edge Detection using Ant Colony Optimization, Color Space-based Thresholding, Genetic Algorithm-based Segmentation and FCM-Based Image Segmentation. In order to quantitatively collate the working of three techniques, the entropy values of skin lesion images are considered. Application of FCM-based Segmentation yields in far better attribute of skin lesion images as compared to Genetic Algorithm-based Segmentation, Edge Detection using Ant Colony Optimization and Color Space-based Thresholding.