Re: Trip Report
CTSC Imaging Informatics subgroup - Nov 7, 2008 mtg in Chicago
- In attendance: Brad Erickson, Dave Channin, Charles Guttman, Bill Hanlon, Kris Juluru, Steve Langer, Dan Marcus, Tony Pan, Fred Prior, Brian Reynolds, Dan Rubin, George Shih, Dan Sullivan
The NIH has asked the CTSC to submit a grant proposal for supplemental funding to create an infrastructure for image data management and sharing among CTSC institutions (PHS 398 format). The proposed system should be a first phase of a working system, not a demonstration project. No budget limit has been specified. Five to seven sites should be included in the first phase. The objectives of this meeting are to determine requirements and scope of the project and to reach consensus on an architecture.
Three domains of potential use are forseen; 1) Clinical, 2) Multi-center clinical trials, 3) Research. To bound the project and help assure success, this proposal will confine itself to research use only, noting that the other domains can be included in future phases.
Required system features:
- Registry of studies (central or decentralized)
- Virtual EMPI
- User authentication and credentialing
- De-identified data only will be captured (mapping to pt identifier is each insitution's responsibility)
- Image annotation
- Workflow: data import and export
- Key Object Selection and annotation
- Admin and Management Reporting
- Adherence to standards (IHE, CDA, DICOM, etc)
- All DICOM data elements of image studies will be indexed in registry
- Support for other metadata e.g. PDFs, clinical reports (I2B2 could be useful here)
Two competing architectural models for image management and sharing are identified; caBIG and IHE. After long discussion of the merits of each, the consensus is that either model could do the job, but that the caBIG would be more acceptable to NIH.
The CTSC Imaging Informatics group will draft a supplemental grant proposal to develop an image data managment and sharing system that fulfills the collaborative requirements of the CTSC institutions. The proposed system will adhere to published standards and include previously developed or NIH-funded Free and Open Source Software (FOSS) as much as possible.
The core architecture of the proposed system will be based on the caBIG infrastructure and include components of the Clinical Trials Processor (CTP), Dorian user authentication and the DCM4CHEE Dicom archive. Other FOSS components such as the Extensible Neuroimaging Archive Toolkit (XNAT), National Cancer Imaging Archive (NCIA) and Annotation and Image Markup system (AIM) will either be included in the core proposal or will be supported if adopted within an individual insititution.
A two-year development time-frame is expected. At least three Use Cases will be featured: Polytrauma, MS and Liver Lesions.
For next step, Brad Erickson will generate a first draft of the "Aims" section of grant proposal and circulate for comment.