Skin Lesion Detection

Abstract

In recent times, various imaging methods and deep learning models have been utilized for identification and analyzation of pigmented lesion images. Clinical and pathological methods of recognizing skin tumors are difficult, time consuming, painful, and expensive. In order to address this problem, many computers aided systems were developed and they achieved great success in detecting several lesions. Owing to the complex behavior of skin lesion images the identification of lesions is still challenging. The identification of skin cancer is making major advances by using the improved CAD models. This study presents an asystematic review of the advances made in each step of a CAD based on deep learning. These steps include pre-processing, segmenting, extracting features, classification, and the state of art approaches used in them. The existing models and the publicly available databases that involve both macroscopic and dermoscopic images are also discussed.