CTSC Brad Dickerson, MGH
The first two steps in the retrospective OFRR1 (Object Free Recall and Recognition) use-case include Basic Data Management and Query Formulation. (Eventually want to upload two projects with matching subjectIDs).
- Step 1: Data Management Describe and upload retrospective OFRR1 datasets (roughly 1 terabyte) and confirm appropriate organization and naming scheme via web GUI.
- Step 2: Query Formulation
- make specific queries using XNAT web services,
- ensure that query results match those returned by analogous searches on the local file system.
- sites involved: MGH NMR center
- number of users: >10
- PI: Brad Dickerson
- staff: Akram Bakkour & Mike Brickhouse (Study coordinator/MRI technician/Lab manager)
- IT staff: NMR could provide IT support
Step 1: Data Management
- Visual confirmation (via web GUI) that all data is present, organized and named appropriately
- How quickly can data of interest be identified via web GUI
- Can data be successfully & efficiently downloaded from XNAT to local file system.
- How does network interruption affect query/download process.
Step 2: Query Formulation
- Successful tests that responses to XNAT queries (for all subjectIDs within some age range and with an MR with particular resolution) results returned from search on the local file system.
- enough documentation to support novice users / developers: as Mike Brickhouse comes up to speed, he will provide feedback about what was missing or challenging in the existing XNAT documentation.
The Dickerson lab has about 250 scans for their own studies, and 150 scans for collaborators. Scans include both MRI and fMRI. Other data are stored with Excel spreadsheets.
- Imaging Modalities
For this OFRR1 use-case, approximately 80 sessions comprise the retrospective data. Data consists of
- MRI (DICOM, .mgz, NIFTI)
- Clinical and Behavioral data ascii text (.txt)
- fMRI paradigms (eprime)
- Some data may contain protected medical information
- Storage needs
Images files are currently distributed over several file servers on the same network.
Current Data Management Process
DICOM files are transferred from the scanners (PACS?) to a local DICOM Server, and then copied to lab machines Images are processed with FreeSurfer.
- Other software to be integrated (e.g., Osirix)
- other features
Target Data Management Process (Step 1.)
(schematic to come)
- Data from multiple file systems will be described and uploaded via an upload script or a light client into a single OFRR1 project on XNAT;
Target Query Formulation (Step 2.)
Step 2. Develop Query capabilities using scripted client calls to XNAT web services, such as:
Show all subjectIDs with age > 60 & MR with resolution=0.38mm Show all subjectIDs with age < 20 & MR with resolution=0.38mm
- Useful product would be a list (csv file?) of appropriate subject IDs, or
- Useful product would be a list of URIs of appropriate datasets
There have been at least two serious attempts using XNAT, working with XNAT support group (NRG) at Harvard. An XNAT with customized schema was developed by Eli White a couple of years ago on Dory at NMR.
The lab recognizes the need for an XNAT-based system to manage images and other data.