Analysis of Skin Lesions for Cancer Detection Using Convolutional Neural Networks

Cancer was Poland’s second leading cause of death in 2021. The average five-year survival rate of patients with skin cancer in Poland is around 70%, much lower than in other countries due to lower early recognition of skin lesions. The computer-aided detection of skin lesions is a relatively new research area that shows the potential to lower the costs of skin cancer screening. Machine learning and deep learning approaches have been successfully used in research on skin lesion detection and classification, which may lead to better-quality treatment based on more detailed and accurate information. The aim of the study was to develop and verify a model of a deep artificial neural network that would indicate the category of skin lesion recognized in the photograph. The convolutional neural network ResNet 152 v2 was used in research. The model developed in this paper, while achieving an accuracy of 86%, exceeds the accuracy of the human eye by 9%.