Deep Learning-Based Identification of Skin Cancer on Any Suspicious Lesion

Abstract

Certain estimates place skin cancer as the most lethal form of the disease worldwide. Spreading to other parts of the body and requiring invasive procedures like chemotherapy and radiation therapy are the results of delayed diagnosis. As a result, automatic methods have greatly aided medical professionals and allowed even non-specialists to identify the sort of cancer a patient has. In this paper, we present a deep learning-based skin cancer identification to classify images of skin lesions as malignant or benign. This method may be used for any potentially malignant lesion. We found that using a deep learning (DL) approach that relied on a mask region-based convolutional neural network yielded the fastest and most accurate results for diagnosing skin cancer.