From Technical Prerequisites to Improved Care: Distributed Edge AI for Tomographic Imaging

Recent years have seen a surge in AI-driven medical image processing, leading to significant improvements in diagnostic performance. However, medical imaging technologies tend to create staggering volumes of medical data, necessitating high-performance computing. Cloud systems with robust GPUs and resource capacity are optimal choices for DL-based medical image processing. However, transferring data to the cloud for processing strains communication links, introduces high communication latency, and raises privacy and security concerns. Consequently, despite the undisputed benefits of cloud computing, dedicated standalone local computers are still used for image reconstruction in today's systems. This localized strategy uses expensive hardware inefficiently and falls short of scalability and maintainability. Edge computing emerges as an innovative concept by bringing cloud processing capabilities closer to data sources. A continuum of computing including local, edge, and cloud tiers would offer a promising solution for medical image processing. According to literature survey, there are no significant works on utilizing edge cloud continuum for CBCT imaging. To fill this gap, we introduce novel 3-TECC architectural concept, specifically designed for CBCT data reconstruction in medical imaging. This article explores the evolving synergy among medical imaging, distributed AI, containerized solutions, and edge-cloud continuum technologies, highlighting their clinical implications and illuminating the potential for transformative patient care. We uncover challenges and opportunities this convergence provides with the CBCT image reconstruction use case, while aligning with regulatory compliance. The proposed 3-TECC architecture advocates a decentralized data processing paradigm, reducing reliance on the centralized approach and emphasizing the role of local-edge computing.