A blockchain-based non-fungible tokens framework for ownership and intellectual property preserving with controlled access to medical datasets

High-quality medical datasets are essential for data-driven healthcare and biomedical research, yet concerns regarding ownership, controlled access, confidentiality, and accountability can limit their sharing and reuse. Conventional repositories provide access management but offer limited cryptographic evidence of dataset provenance and limited mechanisms for tracing post-access redistribution. This study proposes MedNFT-Gov, a blockchain-based governance framework in which each medical dataset is represented by a Non-Fungible Token (NFT), while sensitive content remains encrypted and stored off-chain. A Merkle Root computed from the encrypted dataset provides a cryptographic integrity commitment associated with the NFT metadata. The proposed architecture separates ownership representation from licensed access and specifies time-limited, identity-bound cryptographic access mechanisms. Digital watermarking, perceptual hashing, and blockchain audit records are incorporated as complementary mechanisms for post-distribution accountability, while a guardian-based recovery mechanism is specified for operational key recovery. MedNFT-Gov was evaluated using the OpenNeuro DS007328 neuroimaging dataset. For 3,151 dataset files, SHA-256 hashing required 0.365 ± 0.006 s and Merkle-tree construction required 0.0073 ± 0.0006 s across 10 runs. Merkle-proof generation and verification required 0.0131 ± 0.001 ms and 0.0486 ± 0.002 ms, respectively. Ephemeral key derivation required 0.010 ± 0.002 ms, while AES-256 decryption required 0.33 ± 0.05 ms per file for the evaluated 20-file workload. Hashing exhibited near-linear scaling over the tested dataset sizes, with R² = 0.992. The watermark evaluation showed high extraction accuracy under several compression, noise, filtering, and rescaling conditions but substantial degradation under cropping and rotation. A minimal NFT registration contract was also deployed on the Polygon Amoy testnet; contract deployment consumed 417,023 gas and dataset minting consumed 56,662 gas. The results provide implementation-level evidence for the computational feasibility of the evaluated cryptographic components and proof-of-feasibility for the implemented NFT registration and ownership-anchoring workflow. They do not establish end-to-end blockchain performance or complete production readiness, because access-request approval, delegated licensing, revocation, and guardian-assisted recovery were not fully implemented and benchmarked on-chain. MedNFT-Gov should therefore be interpreted as an integrated architectural approach for medical data governance rather than as a fully deployed operational platform. Its residual limitations include the reactive nature of post-distribution attribution, vulnerability of the evaluated watermark to geometric desynchronization, dependence on auxiliary key-management infrastructure, and jurisdiction-dependent legal recognition of NFT-based ownership claims.