Performance Analysis on Dermoscopic Images for Enhancing the Diagnostic Support System in Healthcare

Abstract

Diagnostic support system plays a significant role in detecting skin cancer. Skin cancers cases are often misdiagnosed due to lack of experience and knowledge of different skin lesion types. Exploring the role of image processing in computer aided diagnosis system. To detect quantification of vascular structures towards skin lesions diagnosis. By taking a dermoscopic image split out the formation of the lesion by initially using independent component analysis into melanin and haemoglobin regions. Haemoglobin component is then clustered into three regions. Shaping filters are also used to erythema region for finding various measures. Hence, a vessel mask is created from global thresholding.12 vascular features are extracted towards lesion diagnosis fed into a classifier and achieves 94.1% accuracy. The proposed technique provides better accuracy and well suited for decision making in diagnostic support system.