Skin cancer is notorious with high numbers of morbidity and mortality. It manifests itself in different layers of epidermis. Melanoma begins in the cells which are in the lowest layer of the epidermis. These cells are known as melanocytes, that is why, melanoma is called as melanocytic skin lesion. Melanoma and nevus are known as melanocytic skin lesions. The global incidence of melanoma in 2019 was 351 880 cases. It was also responsible for 59 782 global deaths. When diagnosed in its earliest stages, Melonoma is curable. This makes the early and accurate detection of melanoma imperative.

Deep CNN(Convolutional Neural Networks) have led to breakthrough in improving the speed and accuracy of melanoma diagnosis. Digital images of skin lesions can be used to educate automated diagnosis systems.

The other type of skin cancer which forms in the upper and middle layers of the epidermis is non-melanoma. It is called non-melanocytic skin lesion. Non-melanocytic skin lesions can further be divided in malignant and benign groups. Non-melanocytic malignant lesions are basal cell carcinoma, actinic keratosis. Non-melanocytic benign lesions are dermatofibroma, vascular lesion and benign keratosis lesions.

Owing to this diversity in non-melanocytic classes, accurate clinical diagnosis of it poses more difficulties, compared to melanocytic cases. This project has shown that with the increment in the sizes of available training sets CNN offers a potential of success in non-melanocytic lesions as well. detecting non-melanocytic lesions. Deep Learning Networks (DLNs) constitute the state-of the art approach for skin cancer screening. DLN´s have posed the potential to produce a seismic shift in skin cancer diagnosis.