Cloud-Enabled Diagnostic System: Leveraging PACS and AI for Enhanced Remote Carotid Atherosclerosis Management

Accurate assessment of the carotid artery vessel wall is essential in managing atherosclerosis. This study presents a cloud-enabled diagnostic system integrating an advanced deep-learning method for automatic segmentation and classification of carotid atherosclerosis in three-dimensional MRI scans. Our solution features a comprehensive software architecture, incorporating a PACS and an enhanced PACS-Viewer with AI-based image analysis and 3D rendering capabilities, aimed at streamlining remote patient management and diagnostic processes. The platform's core is an attention U-Net model, achieving a DICE coefficient of 0.86, and an ensemble of CNN, CBAM, and a pre-trained inception model, yielding 89% accuracy in classifying disease severity. These results highlight the system's precision in segmenting the carotid artery and assessing atherosclerotic conditions. Additionally, our architecture facilitates secure data exchange and storage, upholds patient confidentiality through role-based access, and enables ubiquitous access via a cloud server, significantly enhancing remote diagnostics and collaboration opportunities. The platform's efficacy in deployment illustrates the potential to revolutionize patient management, particularly in resource-constrained environments.