遠隔医療アプリケーションにおける電子患者情報セキュリティのための可逆データ隠蔽

Telemedicine workflows require the secure transmission of medical images while preserving diagnostic integrity. We propose a Reversible Data Hiding in Encrypted Images (RDH‑EI) scheme tailored for telemedicine that couples (i) a Generation of Encryption Parameters (GEP) mechanism to derive per‑block embedding controls from a data‑hiding key and (ii) a two‑level Least Significant Bit (LSB) strategy that provides separable payload extraction and exact image recovery. Images are first encrypted using AES‑CTR with a unique nonce; GEP then produces block‑wise traversal orders, offsets, and parity masks used by Phase‑1 (odd blocks, parity‑of‑triples) and Phase‑2 (even blocks, serialized side‑information). The method supports three operating modes: payload‑only extraction, decrypt‑only viewing, and full, bit‑exact recovery when both keys are available. On a 90‑image test set (30 X‑rays, 30 MRIs, 30 CTs, all 512 × 512), the proposed approach achieves higher quality on directly decrypted images than representative baselines at Z = 16: PSNR 39.92 ± 0.41 dB (X‑ray), 37.78 ± 0.38 dB (MRI), and 38.27 ± 0.36 dB (CT), with SSIM 0.9784 ± 0.0021, 0.9694 ± 0.0023, and 0.9835 ± 0.0018, respectively. The recovered images are bit‑exact (PSNR = ∞). Encryption robustness is supported by near‑maximal entropy, high NPCR (> 99%), and UACI near 33%. These results indicate that the proposed GEP‑driven, two‑level embedding improves the payload-distortion trade‑off while retaining strict reversibility, making it suitable for secure medical‑image sharing in telemedicine.