<![CDATA[Advances in artificial intelligence have given rise to challenges involving realistic deepfake images, extending into the healthcare sector. The contribution of this study lies in the implementation and performance analysis of YOLOv11 for detecting medical image deepfakes on a lung CT scan dataset covering variations of benign and malignant cases. The scope of the study is limited to binary classification between authentic and fake images, tested in a staged manner. CT-GAN and stable diffusion (SD) manipulation methods are employed to evaluate model performance. The results show that the YOLOv11 model achieves 100% accuracy, precision, recall, and F1-score on images manipulated using stable diffusion. In contrast, CT-GAN–based manipulations present challenges in distinguishing between authentic and fake lung cancer CT scan images. With further improvements and enhancements, fine-tuned YOLOv11 has the potential to become a relatively lightweight, fast, and accurate model for medical image deepfake detection. These results have the potential to support patient data security and maintain the integrity of clinical diagnostics in the future.]]>
Copyrights © 2026
