A Hybrid Blockchain and Convolutional Neural Network for Secure Cloud-Assisted Medical Image Analysis

In recent years, securing medical images in the cloud has become essential owing to the rapid accumulation of
Computed Tomography (CT), Magnetic Resonance Imaging
(MRI), and ultrasound data from hospitals, diagnostic centers, and IoMT-enabled devices. However, the Customizable Unique Node Access (CUNA)-based deep encryption model has limited scalability, verifiable provenance, and lacked advanced analytics for accurate medical interpretation. To overcome these limitations, this research proposes a hybrid Blockchain and Convolutional Neural Network (B-CNN) model for secure cloud-assisted medical image processing. Initially, data is collected from medical imaging devices integrated with hospital information systems and sensors. Preprocessing at the edge nodes then performs normalization of image dimensions, anonymization of patient identifiers, denoising, and metadata generation. Encrypted medical data are stored off-chain, whereas a permissioned blockchain records immutable hashes, access policies, and consent. A CNN–transformer hybrid network extracts spatial features through convolutional layers and global context from transformer encoders and then combines them for accurate classification to enhance scalability, security, diagnostic precision, and trust in cloud healthcare systems. The experimental results demonstrated that the proposed B-CNN model outperformed the existing CUNA
model in terms of accuracy (93.23%).