Semantic segmentation-based skin cancer detection

Melanoma detection requires dermoscopy image segmentation, one of the medical image segmentation domains. Melanoma, the most dangerous form of skin cancer, can strike without warning on healthy skin or grow from a pre-existing lesion. Skin lesion border segmentation is required to find lesion locations in histopathologic imaging effectively. True, accurate skin lesion segmentation remains challenging due to blurring boundaries, necessitating an accurate and automated skin lesion segmentation approach. Semantic segmentation using deep learning architecture has been effectively employed in this paper for the desired task. The segmentation of the skin lesson is performed using u-net-based deep learning model and analyzed using standard and valid evaluation metrics. The simulation is carried out on the available dataset and implemented on the Python Collab (R) tool.