Melanoma Detection Using Convolutional Neural Network

Melanoma is a fatal form of skin cancer that needs to be detected as early as possible for life-saving and better survival. The existing method is the manual examination which might lack accuracy or lead to misdiagnosis with respect to the training and experience of the dermatologist, who detects if it is melanoma or other skin disease using dermoscopy, will be very time-consuming while the patient goes through much suffering hence we are pushed to the scope of computer-assisted technique where in deep learning technology has a major role nowadays. This paper will evolve around the hybrid combination of deep learning techniques such as Convolutional neural network, encoder-decoder network, and Recurrent Neural network for finding the best accuracy and sensitivity of the lesion from the datasets taken from the ISIC archive (International Skin Imaging Collaboration). The images are pre-processed, and segmented using pixel-based—ANN and fuzzy c-means clustering, Features are extracted, and finally the results are classified in a relatively faster time to help doctors treat the patient more efficiently.