Abstract
Skin cancer is one of the most common types of cancer in terms of incidence around the world. Melanoma, in particular, manifests as visible pigmented skin lesions, often a result of sun overexposure. Despite the fact of rapid development of dermatology, delays in diagnosing skin diseases and poor access to their treatment are occurring in modern medicine.Several studies have demonstrated the potential of CNN in developing accurate and noninvasive screening methods for skin cancer. Nowadays, a limited number of mobile applications are identified that involve deep learning for the scanning and screening of skin diseases.This study simultaneously proposes a CNN architecture with a non-standard activation function, as well as presents the results of the proposed CNN model in terms of various metrics, such as accuracy, recall, specificity, precision, and so on. The software development is implemented in the Flutter framework, focusing on the Waterfall model. The key components of the mobile application, including the splash screen, main page, navigation, health-related features, and the "About Us" section, are described.The results of this study demonstrate that the proposed CNN model achieved 81.53% in terms of accuracy. The image augmentation techniques employed contribute to improved model generalization and robustness. By combining the capabilities of CNNs with user-friendly mobile interfaces, individuals can gain easy access to melanoma assessment, promoting proactive healthcare practices.