Skin Lesion Analysis Towards Melanoma Detection Using Deep Learning Network

Skin lesion is a severe disease in world-wide extent. Reliable automatic detection of skin tumors is very useful to help increase the accuracy and efficiency of pathologists. International Skin Imaging Collaboration (ISIC) is a challenge focusing on the automatic analysis of skin lesion. In this brief paper, we introduce two deep learning methods to address all the three tasks announced in ISIC 2017, i.e. lesion segmentation (task 1), lesion dermoscopic feature extraction (task 2) and lesion classification (task 3). A fully-convolutional network is proposed to simultaneously address the tasks of lesion segmentation and classification and a straight-forward CNN is proposed for the dermoscopic feature extraction task. Experimental results on the validation set show promising accuracies, i.e. 75.1% for task 1, 84.4% for task 2 and 90.8% for task 3 were achieved.