Convolutional Neural Network Based Skin Lesion Analysis for Classifying Melanoma

In a human body skin is the core part, which helps to cover the muscles, bones what's more with the entire body. These days numerous people are suffering from skin diseases. Melanoma is the enormous type of skin malignant growth and the extent of these skin diseases is increasing day by day. Recognizing the type of skin disease automatically from the images can assist in the quick diagnosis and enhanced accuracy saving valuable time. Here, a machine learning based technique using convolutional neural network (CNN) for classifying seven types of skin diseases has been proposed. Transfer learning, along with CNN, has been used to improve the classification accuracy on the International Skin Imaging Collaboration 2018 (ISIC) dataset. Evidence of 11% increase in the accuracy by using transfer learning than using only CNN has been found. Compared to some existing works, performance of this proposed method is promising.