Enhancing dermoscopy images to improve melanoma detection

Since the application of image analysis algorithms has received much interest in the field of medical imaging, particularly to aid physician with their diagnosis, ways of improving the analysis accuracy in this field has become an important research question. In this work we investigate into the performance analysis of an automatic lesion classification algorithm proposed for skin cancer detection. We show that enhancement and color correction of the original lesion images have a positive impact on the further cancer detection, in the process of automatic classification and visual judgment. The purpose of the work is to analyze the influence of different image enhancement and color correction methods on efficiency of melanoma detection results, tested on a given set of dermoscopy images and a previously developed CNN-based detection algorithm.