Medical Image Tampering Detection using Deep Learning

The increasing prevalence of cyber-attacks on hospitals and the emergence of advanced photo-editing tools have raised concerns about the potential for medical image falsification. This research focuses on detecting and preventing medical image malpractice, particularly in the context of computed tomography (CT) scans. Building upon previous work with VGG19 and ConnectionNet, this study explores the potential of state-of-the-art deep neural networks, such as ResNet50 and MobileNetV2, for advanced medical image identification and prediction. These models are known for their ability in image classification and can achieve exceptional results when trained on large datasets. By analyzing a diverse dataset of fabricated CT images, this research aims to extract the characteristics of various forgery techniques. The proposed approach seeks to improve accuracy, scalability, and versatility in medical image malpractice detection. This study contributes to preserving the reliability of diagnostic information and ensuring the integrity of medical imaging, safeguarding patient care and preventing misdiagnosis.