A Differential Evolution approach for classification of Multiple Sclerosis lesions

The problem of automatically extracting novel and interesting knowledge from large amount of data is often performed heuristically when pattern extraction through classical statistical methods is found hard. In this paper an evolutionary approach, based on Differential Evolution, is proposed, which is able to perform the automatic discovery of comprehensible classification rules as a set of IF?THEN rules over a database of Multiple Sclerosis potential lesions. Moreover, this tool also determines which the most discriminant database attributes are in categorizing instances. Therefore, this evolutionary tool provides an efficient decision support system for clinical decisions, that could be a useful tool for medical experts to help them gain insight into the reasons for assessing the abnormality of a lesion.