Convolutional Neural Networks Based Study and Application for Multicategory Skin Cancer Detection

Skin cancer is a very common form of cancer that jeopardizes people's health. Like other cancers, early detection is crucial to its treatment. However, traditional methods for diagnosing skin cancer can be low in accuracy and often leads to unnecessary examination. In addition, some existing Machine Learning models for skin cancer detection can be limited as well for the small number of skin cancer categories they support. In this work, three types of Convolutional Neural Network (CNN) models are compared on a nine-class skin cancer classification task and the model with the highest accuracy is integrated into a web application. The three CNN models that are compared include VGG-16, VGG-19, and a self-designed network. Since the three models differ in their depth, the relationship between the depth and performance of a model within the scope of the dataset used was also explored. Test results showed that the most accurate model is VGG-19, which achieved 0.9290 in accuracy and 1.2842 in loss, making it a reliable method to assist skin cancer detection.