Color calibration model of skin lesion images for computer-aided diagnostic

Computer-aided diagnostic is developing rapidly and is being introduced in multiple fields of medicine. One of those fields is dermatology where various types of equipment are used for image acquisition. Those images can be processed in order to assist diagnostic by giving comparative examples, initial image enhancement or automated evaluation. In our research we concentrate on creation of computer-aided methods for skin lesion diagnostics, especially Malignant Melanoma screening based on ABCD evaluation rule [1]. While preparing these methods a number of issues appeared over and over again. The problems are due to the fact that we work with datasets that were obtained from equipment with different characteristics. The methods giving excellent results on one set of images may not perform as well on the others. Solving this problem usually requires manual fine-tuning of a set of parameters. In this paper we introduce a method of automated image adjustment based on color calibration model for calculating the Total Dermoscopy Score (TDS) of the ABCD rule. First we describe the general problem of skin lesion diagnostic and our motivation for the research. Next we will introduce the idea behind the method and its implementation. Finally the practical usage will be shown and discussed.