Our project and our vision

The open medical inference (OMI) methods platform will enable the discovery and use of remote AI services. OMI will specify open protocols and data formats for the semantically interoperable peer-to-peer exchange of multimodal healthcare data and remote AI inference. We will establish an initial selection of services, focusing on image-based multimodal AI models.

OMI’s open protocol for data exchange will build on the data sharing common framework across Medical Informatics Initaitive (MII) consortia. To maximize interoperability, we will actively participate in the MII WG Interoperability (WG IOP), specifically in the development of the specification and implementation guideline for the medical imaging extension module of the MII core data set. With OMI, we will establish a link between FHIR and DICOM via FHIR endpoint definitions of DICOMwebTM-capable DICOM nodes.

OMI will provide a generic open-source gateway component that enables RESTful access to legacy PACS at all partner DICs via a subset of the DICOMwebTM API specification. OMI components include a gateway server to connect AI services to the MII DSF, a client to enable DICs and data management service providers to access OMI gateway servers, and a service registry to discover and check the status of connected AI services.

We will ensure the seamless integration of OMI with the MII by a) integrating existing MII structures and concepts b) using local MII data integration center components (e.g. pseudonymization services and consent management) and c) using open standards while focusing on simple, modern, and common technologies such as REST, TLS, FHIR, and DICOMwebTM. This design will keep entry-barriers at a minimum. Our project partners will establish a network of service recipients and service providers in the final project phase. We will test the functionality, security, and usability of the OMI specification and reference architecture.

OMI is one of the MII use cases in the extension phase and is funded by the German Federal Ministry of Education and Research (BMBF) with more than 8 million euros from 01.07.2023 to 30.06.2027. In this cross-consortium project OMI, 16 partners from the four medical informatics consortia DIFUTURE, SMITH, HiGHmed and MIRACUM are working together to establish a network of users and providers of AI models to simplify the use of artificial intelligence in performing time-consuming and repetitive tasks in medicine. The project is coordinated by the University Medical Center Essen.