Developing a Hybrid Deep Learning Framework for Automated Skin Cancer Classification Using ISIC Dataset

Abstract

Skin cancer is one of the worst kinds of cancer, skin cancer has a high death rate if it is not identified and treated quickly. One of the main causes of this condition is the fast proliferation of skin cells brought on by exposure to sunshine. Reliable automated solutions for wound recognition must be developed because early detection is essential for successful treatment. Using photos from the ISIC dataset, this study suggests an automated method for identifying skin cancer by utilizing mixed algorithms like Xception, as and ResNet50 and deep learning techniques like transfer learning. This strategy seeks to lessen the workload for medical professionals by creating a web application with Stream lit, enabling them to take preventative action as soon as possible. The system's primary goal is to categories different forms of skin cancer, such as actinic. The approach is designed to categories different kinds of skin cancer, such as melanoma, carcinoma of basal cells, dermatofibroma, and actinic keratosis. An important step forward in the use of AI in healthcare is represented by the suggested solution.