SECURED AI-ENABLED FRAMEWORK FOR MRI SCANNER HEALTH MONITORING USING XG BOOST-BASED FAULT DIAGNOSIS AND STEGANOGRAPHY IN IOMT

Abstract

Reliable health monitoring of magnetic resonance imaging (MRI) scanners is essential in modern healthcare environments supported by the Internet of Medical Things (IoMT) to ensure uninterrupted clinical services, minimize downtime, and facilitate predictive maintenance. However, the continuous transmission of maintenance and diagnostic information over IoMT networks exposes sensitive data to security threats, including interception, tampering, and integrity attacks, which may result in inaccurate fault diagnosis and inappropriate maintenance decisions. The primary objective of this work is to develop a secure and intelligent MRI scanner health monitoring framework that integrates machine learning (ML)-based fault classification with reversible steganography, tamper detection, localization, and self-recovery mechanisms to ensure both diagnostic accuracy and secure data transmission in IoMT environments. The proposed consensus-based distributed recovery steganography for IoMT (CDRS-IoMT) framework utilizes MRI scanner maintenance records, including device age, maintenance cost, downtime, maintenance frequency, failure event count, and other device-related attributes, to classify maintenance conditions using an Extreme Gradient Boosting (XGBoost) model. The predicted maintenance class, confidence score, timestamp, and authentication metadata are securely embedded into monitoring images through reversible data hiding techniques based on Least Significant Bit (LSB) matching and prediction-based embedding. Furthermore, a consensus-driven distributed recovery mechanism stores recovery information across spatially separated image blocks to improve resilience against localized tampering. At the receiver side, hash-based authentication is employed for integrity verification, followed by tamper localization and image reconstruction when modifications are detected. Experimental results demonstrate that the proposed CDRS-IoMT framework achieves a classification accuracy of 98.7%, precision of 98.1%, recall of 97.9%, and F1-score of 98.0% for MRI scanner maintenance classification. Additionally, the framework attains an embedding peak signal-to-noise ratio (E-PSNR) of 58.42 dB, embedding structural similarity index measure (E-SSIM) of 0.9996, recovery PSNR (R-PSNR) of 28.41 dB, and tamper localization accuracy (TLA) of 98.2%, demonstrating excellent visual quality, robustness, and reliable tamper detection and recovery. The proposed work provides a secure, intelligent, and resilient solution for IoMT-enabled MRI scanner health monitoring by combining AI-based maintenance classification with reversible steganography, thereby preserving the confidentiality, integrity, and authenticity of maintenance information while supporting trustworthy predictive maintenance in smart healthcare environments.