A Refined Approach for Classification and Detection of Melanoma Skin Cancer using Deep Neural Network

Although being a less common form of skin cancer, melanoma is the deadliest of all, accounting for around three-quarters of skin cancer-related deaths. The epidemiological learnings at hand clearly show the relationship between solar UV radiations and skin cancer. For curing it on time, the early-stage identification of melanoma is very necessary. Depending on the clinical aspects of melanoma, appropriate microscopic (dermoscopic) and macroscopic (clinical) analysis are enacted to detect the malignant melanoma. Digital image classification of skin lesions is the basis of efficient skin cancer diagnosis that reduces the time spent and pain received by victims in detecting early melanoma. In this paper, we introduced computer supported strategies for the recognition of melanoma skin cancer utilizing multiple image processing tools. Firstly, a skin lesion image acts as an input to this system and then various image processing and classification agents deduce the presence of melanoma. These analysis techniques test for the melanoma warning signs like border, color, size, and shape for segmentation and feature extraction. This piece of work describes the various approaches of image processing to have improved diagnosis of melanoma skin cancer.