This project plans to develop an advanced system for detecting skin cancer by analyzing lesion images, utilizing the ISIC 2019 and HAM10000 datasets. These datasets encompass various skin lesion images, ensuring a comprehensive representation of different skin cancer types, which is crucial for training a robust and accurate model. The project innovatively combines incremental learning and federated learning techniques to create a model that is adaptable and prioritizes user privacy. Incremental learning allows the model to continuously evolve as new data becomes available, ensuring it remains current and effective in recognizing emerging patterns and variations in skin cancer presentations. Meanwhile, federated learning enables the model to learn from data distributed across multiple sources without transferring sensitive information to a central server, enhancing data security and protecting patient confidentiality. By leveraging the diverse samples within the ISIC 2019 and HAM10000 datasets, this approach aims to improve the model's ability to accurately classify various skin cancer types, ultimately advancing the application of artificial intelligence in medical diagnostics. The emphasis on security and adaptability aligns with current healthcare data regulations. It fosters greater trust among users and medical professionals, paving the way for more widespread adoption of AI-driven diagnostic tools in clinical settings.
| Name | Student Number |
|---|---|
| Bekir Emrehan ŞİMŞEK | 202011039 |
| Bilgesu FINDIK | 202011407 |
| Melike Hazal ÖZCAN | 202011013 |
| Pelin KOZ | 202011048 |
Dr. Ayşe Nurdan SARAN