Melanoma Classification Using Deep Learning

Abstract

Melanoma is a life-threatening and serious skin cancer with an increasing number of fatal cases worldwide. Early detection of Melanoma is of vital importance because the survival of the patient is based on the stage of the disease. In the first stage survival rate is more than compared to further stages. Melanoma is a life-threatening and serious skin cancer type with an increasing number of fatal cases worldwide, early detection of Melanoma is of vital importance because of its survival rates at different stages that is starts from 1st stage and 98.4% at 5th stage. With the advancement in technology, early detection is possible through several applications which uses the concepts of machine learning and deep learning. In this paper, a convolution deep learning model named MelaNet is designed from scratch to classify benign and malignant type of melanoma skin cancer. MelaNet achieved the high accuracy of 92.67% to classify the skin lesion accurately. One of the main differences between MelaNet and other existing models is that MelaNet's parameters are optimally specified and tuned, which aids in avoiding over/under-fitting.