MediMarkChain for blockchain anchored medical image provenance using DWT SVD watermarking and CID linked certificates
Year: 2026
Type: article
Abstract: Medical image exchange requires reliable integrity checking and verifiable provenance, whereas many existing approaches address watermark robustness or auditable registration in isolation. MediMarkChain integrates DWT–SVD watermark embedding, Reed–Solomon-protected identity payloads, CID-linked off-chain provenance artifacts, and a blockchain registry to provide strict tamper evidence for medical images. The payload is generated from a SHA-256 watermark identity and protected using Reed–Solomon parity. The registry records the issuer address, timestamp, hashes of the original image, protected image, and payload, algorithm version, and CID reference. Evaluation on a balanced public subset of 320 chest X-ray images covers dataset-wide image quality, geometric and combined attacks, subband comparison, parameter ablation, and registry overhead. Under the baseline single-level Haar DWT configuration with LL -subband embedding, the protected images achieved a mean PSNR of $$51.64\pm 0.31$$ dB, SSIM of $$0.99865\pm 0.00036$$ , VIFP of $$0.9928\pm 0.0020$$ , and UQI of $$0.9979\pm 0.0032$$ . Among the tested single-subband configurations, LL embedding provided the strongest combined PSNR, SSIM, and VIFP performance. Smart-contract registration required $$207161\pm 4$$ gas per record and achieved a mean local registration latency of 29.2 ms. Exact protected-image hash verification returned no registered match for any attacked variant, confirming byte-level change detection throughout the evaluated attack suite. However, the current non-blind additive extraction mechanism achieved 0% exact payload recovery on both clean protected images and attacked images. These findings establish MediMarkChain as a reproducible strict-provenance and audit framework for medical images, while transformation-tolerant payload recovery remains an unresolved extension.
Source: Discover Artificial Intelligence
Authors: Ashish Pandey, Madhu Sharma, Chirag Joshi
Institution: Manipal University Jaipur
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,590
OpenAlex ID: W7225647933