AN EFFICIENT DEEP LEARNING BASED SEGMENTATION AND CLASSIFICATION FRAMEWORK FOR SKIN LESION IMAGES

Abstract

Skin lesion segmentation plays a significant part in the earlier and precise
identification of skin cancer using computer aided diagnosis (CAD) models. But, the
segmentation of skin lesions in dermoscopic images is a difficult process due to the
constraints of artefacts (hairs, gel bubbles, ruler markers), unclear boundaries, poor and so on.
This paper presents a new deep learning (DL) based skin lesion segmentation and
classification model. The proposed model involves different stages namely pre-processing,
segmentation, image enhancement, feature extraction and classification. The segmentation
process is carried out using deep instance segmentation (DIS) technique. Then, the images
get enhanced by removing the hair presented in the segmented lesion region. Afterwards, the
features are extracted from the image using Inception ResNet v2 model. Finally, support
vector machine (SVM) model gets executed to classify the images into respective classes. For
examining the effective outcome of the proposed model, a detailed comparative analysis with
earlier models takes place. The simulation results exhibited superior performance of the
presented model under diverse aspects.