Radiology research platform: automating multicenter radiological research workflows using a cloud-native platform.

Large-scale, multi-institutional radiological datasets, requisite for next-generation AI and epidemiological studies, remain an unmet need due to fragmented infrastructure, institutional firewalls, and the laborious manual entry of data into research-specific electronic data capture tools. We present a proof-of-concept, cloud-native platform to facilitate multicenter radiological collaboration. The platform architecture enforces a clean separation between a React-based client UI, a cloud backend in Google Firebase, and secure local networks. The image data and metadata of each institution are securely and locally hosted through a localized edge-node Orthanc DICOM server. Critical metadata are reverse-proxied to the cloud database via REST APIs, negating any need for complicated virtual private network setups. Using dynamic JSON-based structured evaluation forms and enrollment algorithms such as consensus or fixed-volume study designs, PIs define clinical workflows to be dynamically pushed to radiologists, who then evaluate pending cases within a single workspace, including an embedded zero-footprint Open Health Imaging Foundation viewer and a standardized reporting form. KEY POINTS: Question How can radiologists from multiple sites and institutions participate in collaborative research projects without requiring VPN configurations or hardware shipments? Findings A cloud-native platform based on localized edge nodes and a zero-footprint web viewer enabled automated synchronization of imaging metadata and dynamic assignment of evaluation queues, eliminating manual data curation steps and streamlining radiologist workflows. Clinical relevance This application overcomes complex IT barriers and bypasses manual data entry to speed up the generation of radiological datasets and studies. Consequently, it accelerates the development of novel protocols and guidelines for patient care worldwide.