Identification & enhancement of different skin lesion images by segmentation techniques

A skin lesion is a portion of skin with an abnormal growth or presence due to a skin disease. Skin diseases at their early stages can be cured very easily; otherwise they begin to spread to other parts of the body and may be deadly. An early detection of skin disease is essential for such patients. Due to the high cost involved in various dermatology testing procedures for every patient, an automated system is required to give better lesion images and vision to help the doctors to further diagnose and prescribe the correct prescription and medication. In this paper, an innovative approach for automatic identification of skin lesions is proposed. To improve the quality of skin lesion images, Median filtering and Sobel edge detection techniques have been implemented for filtering and segmentation. The efficacy of the proposed work has been verified by measuring the entropy of the resultant images obtained for different skin diseases. The performance is tested on a dataset of 70 samples from 150 medical images of different body parts with 10 different classes of skin diseases.