Melanoma Skin Cancer Detection and Classification using Deep Learning

Abstract

Abstract: Melanoma is caused by abnormal growth of skin cells which is most often devloped on skin exposed to UV rays. It is less common but one of the most dangerous diseases in the world. Classifying skin lesions correctly at an early stage could avoid some types of skin cancers like melanoma and increases the chances of cure before cancer spreads. One of the efficient methods to accurately identify Melanoma is using deep learning. Deep learning methods such as AlexNet, Inception V3 were proposed in the past. In this project, other methods like MobileNet, CNN, ensemble method, ReNet, VGG16 will be used to detect the type of cancer using dermoscopic image data. Layers of these models are applied for fine-tuning which allows us to differentiate between different classes of skin lesions. There are two kinds of Melanoma, malignant and benign and we will also perform grading to early diagnose the disease which will ultimately increase the survival rate.