Automatic Kaposi's sarcoma detection using texture distinctiveness

Abstract

There is a growing emphasis on skin cancer diagnosis and Kaposi's sarcoma has recently received increasing attention. Kaposi's sarcoma is one form of skin cancer. The time and costs required for medical experts to screen all patients for Kaposi's sarcoma are prohibitively expensive. Dermatologists need an automatic diagnosis system to assess a patient's risk of Kaposi's sarcoma without using special or costly equipment. One challenge in implementing such a system is locating the skin lesion. We propose Texture Distinctiveness Lesion Segmentation Algorithm (TDS-KS) to automatically locate skin lesions from the photograph. TDS-KS algorithm consists of two main steps. First a set of representative texture distributions are learned from the input skin lesion image and texture distinctiveness metric is calculated for each distribution. Then a texture-based segmentation algorithm classifies regions the input image as normal skin or lesion based on the occurrence of representative texture distributions. The input images are taken from dermquest database which has images of different skin diseases.