LUCIDNet: Deepfake Generation and Detection in CT-Scan Lung Cancer Imaging via Stable Diffusion and Enhanced EfficientNetV2-L

The application of Artificial Intelligence (AI) in healthcare, particularly in medical imaging, is rapidly advancing, ensuring both the integrity and safety of medical data. This dual-focus approach underscores significant technological advancements, combining the creation of synthetic medical images for research purposes with robust deepfake detection methods to ensure authenticity. This study introduces LUCIDNet, a comprehensive framework that integrates a Stable Diffusion Model with a UNet backbone and attention mechanisms for generating highly realistic synthetic CT scans of lung cancer, achieving an impressive 99.2% image generation accuracy. Additionally, an Enhanced EfficientNetV2-L Model is incorporated for robust deepfake detection, addressing the critical need for verifying the authenticity of medical images. The performance of LUCIDNet is evaluated using metrics such as precision, recall, F1 score, and loss, demonstrating superior performance compared to existing state-of-the-art methods. Our research not only advances the field of synthetic medical image generation but also provides a reliable solution for detecting deepfake medical images, thereby safeguarding patient health and the integrity of medical data. Our code is available at (https://github.com/nani67/LUCIDNet/tree/main).