Tampering detection and localization in medical images using DCT and chaotic encryption

Medical imaging has undergone a sea change due to advancements in medical technologies. In recent years, image sharing through public networks has developed into a widely accepted collaborative modality among medical practitioners. However, transmitting sensitive patient information over such networks introduces significant risks, including unauthorized access, data tampering, and privacy breaches. Traditional approaches for watermarking may be effective in authentication but degrade the quality of the image, possibly leading to a wrong diagnosis. This paper proposes a methodology for medical image integrity verification that ensures secure tamper detection, modification localization, and preservation of image fidelity. Experimental results on various publicly available medical databases and clinical data demonstrate that this method outperforms previous works with an accuracy in detecting subtle attacks of 98.5-99.3%, false positive/negatives as low as 0.8%, near-optimal entropy values of 99.95-99.98%, while achieving a real-time processing latency of less than two seconds.