Medical imaging techniques play a vital role in identifying skin cancer. The detection of skin cancer in its early stage is very crucial and important. This project shows a comparative study of two different segmentation method for segmenting skin cancer region from dermoscopic image. Content-Adaptive Superpixel (CAS) segmentation is based on Clustering and Semantic segmentation is based on Artificial Neural Network (ANN). The goal is to find an efficient method for detection of skin cancer from a dermoscopic image. The proposed model comprise of Preprocessing, Segmentation using CAS and Semantic segmentation. The Fully Convolutional Network and RGB conversion is used for semantic segmentation. CIELAB conversion and modified linear clustering algorithm for CAS segmentation. The experimental results confirm that performance on semantic segmentation is better than CAS.

Keywords : Skin Cancer, Segmentation, Dermoscopic image, Preprocessing, Segmentation, Content-adaptation, KNearest Neighbor Classifier.