Detection of GAN-manipulated Medical Images through Deep Learning Techniques
S. Aruna, Surabhi Narayan · 2024
with advancements in technology, telemedicine has become popular to provide remote clinical services in real-time using digital audio-visual communication means. Medical images like scans and reports are transferred over the internet to facilitate remote diagnosis. While transferring these medical images over various communication channels, medical images are susceptible to deliberate manipulations by attackers. Medical image tampering is the deliberate manipulation of medical images with potential consequences for patient safety, healthcare confidence, and research integrity. The integrity of healthcare institutions, patient safety, and ethical norms are all upheld by forgery detection. In this paper, we propose a framework to enhance the model's resilience against potential manipulations in CT scans, alleviating concerns about deep fake attacks on medical images. The framework integrates L2 Regularization, LBP pre-processing approach, SVM Classifier, and U-Net architecture for efficient detection of forgery. The proposed model achieves a detection accuracy of 93.9%, precision of 94.4%, recall of 93.9%, F1 score of 94% and the area under the ROC curve of the receiver is 99.2%