Intelligent System for Early Diagnosis of Skin Cancer using Bilinear CNN and Soft Attention Approach

Abstract

Skin cancer is becoming the leading cause of death in the Western world. Although sun exposure is a major risk factor for the development of skin cancer, this malignant neoplasm can appear anywhere in the body’s skin. When detected early, most cases of skin cancer can be treated successfully. Skin cancer is fatal if not caught and treated quickly. New tools allow for the first-stage detection of skin cancer. The biopsies are the official method of skin cancer diagnosis. This is accomplished by scraping out a small amount of skin and sending it off to the lab for analysis. It takes a lot of effort and time. The suggested skin cancer detection method makes use of BilinearCNN-SA to identify malignant moles at an early stage. Patients benefit more from it. Preprocessing steps in the diagnostic approach include various elements like noise removal, grayscale conversion, and image enhancement. Once the data has been cleaned up, they apply Otsu segmentation. The ABCD Rule is used for feature extraction. The BilinearCNN-SA Model is then used to make the final classification. When compared to convolutional neural network and support vector machine models, the proposed method fares very well.