Analysis and Classification of Skin Cancer Using Deep Learning Approach

Disease that occurs from the unregulated expansion of irregular epidermal cells is always considered to be deadly and of utmost concern, if not recognized and treated during a preliminary phase. Any sort of melanoma cancer's preexisting diagnosis encourages hope for improvement. But the method employed in diagnosis of such a phenomenon is a laborious procedure. This work proposes a computer assistance based deep learning method to detect and analyze the irregular expansion of the epidermal cells. The method of diagnostic imaging with computer assistance has proven to be more effective in recent past. Deep learning is a kind of artificial intelligence that mimics how the human mind processes information and creates connections to aid in judgment. Convolutional neural networks are being employed in this study to examine and classify different categories of images. The algorithms used for classification and analysis are Inception-V3, Mobile net, Vgg-16, and a custom CNN model from scratch. This work also examines other AI-based skin cancer screening methods as well. In this case, a pre-trained model is reused utilizing the approach of transfer learning, and a new model was also created from start utilizing CNN structures. A website application that uses JavaScript, Flask, and CSS is indeed exhibited; all we need to do is upload the inspection picture, and it will forecast the results. A web app aids in providing the individual with the outcome. In a nutshell, the proposed work would help in earlier diagnosis of skin cancer.