Color and Spatial Features Integrated Normalized Distance for Density Based Border Detection in Dermoscopy Images

Abstract

Dermoscopy is the major imaging modality used in the diagnosis of melanoma and other pigmented skin lesions. Automated assessment tools for dermoscopy images have become an important research field mainly because of inter- and intra-observer inconsistencies at interpretation of the same image. Automated detection of lesion borders is very important step in dermoscopy image analysis. One of the highest accuracy rates in the automated lesion border detection field is achieved by Fast Density Based Lesion Detection (FDBLD), which is based on density based clustering of pixel-of-interest. In addition to low border detection error, FDBLD removes redundant computations in well-known spatial density based clustering algorithm DBSCAN; thus, in turn it accelerates clustering process considerably. However, FDBLD is designed to run only on binary images; thus, it requires pre-processing step to convert color image in to a binary image. Furthermore, very important color information in dermoscopy images falls into disuse.