An Efficient Content‐Based Authentication Scheme for Tamper Detection and Localization With Medical Images Using <scp>ORB</scp> Keypoints and Voronoi Tessellation

ABSTRACT Medical images play a vital role in accurate disease diagnosis; however, the widespread digital transmission and sharing of medical data raise critical concerns regarding image integrity and authenticity. This paper presents an efficient fragile watermarking framework for medical image authentication and tamper detection based on content‐derived image fingerprints. The image is partitioned into non‐overlapping blocks, and a unique, regenerative fingerprint is computed for each block to enable precise tamper localization. A key contribution of the proposed method is the adaptive classification of blocks into Region of Interest (ROI) and Region of Non‐Interest (RONI) based on ORB keypoint density and block‐level analysis. For ROI blocks, Region‐Based Signatures (RBS) are generated using Voronoi tessellation of ORB keypoints with incremental reduction, producing spatially unique and robust fingerprints. The RBS coordinates corresponding to ROI blocks are embedded into RONI blocks using LSB substitution while preserving the original topological structure. The proposed scheme is evaluated on MRI, CT, X‐ray, and MR angiography images. Experimental results demonstrate high imperceptibility and reliable authentication performance. Comparative analysis with existing watermarking techniques further establishes the proposed method's superiority in terms of computational efficiency, visual quality preservation, and tamper localization accuracy.