Mbirn:DTI Action Items Sedona 2006

DTI Action Items Arising from the M-BIRN All Hands

• Recommended cross-platform acquisition parameters for DTI at both 1.5T and 3.0T

The agreed upon goal is to produce BIRN-recommended acquisition parameters and B0- and eddy-current corrections for use in clinical studies. There are several outstanding issues. Anders Dale and Allan Song have different correction algorithms that both should be explored and compared to determine if there is one best way of doing this. The group of Helmer, Mori, Song, Farrell, Dale and consultant Pierpaoli will meet offline to work on acquisition parameters. The raw and processed image data from this action item will will be posted on the BIRN website. This working group will explore QA/QC methods and explore developing these into a workflow pipeline. In collaboration with the FBIRN, the travelling DTI phantom project to assess intersite variability will occur after this work has been finished (in 2007).

• Issues with DTI file formats

1. NRRD format - discussion with NIFTI and MNI people to discuss the adoption of a common format and issues that may have been address by those groups in drafting the BXH standard (Gordon Kindlmann, Syam Gadde, Dave Keator, Steve Pieper). Several outstanding issues to be worked out by this group include:
2. How does one obtain the information needed to transform data into NRRD format? Will BIRN supply a software tool to convert the various DICOM versions to NRRD or will we supply a wiki page to document the files needed to figure it out on your own? (who?)
3. The BIRN Ontology group will work together with these domain scientists and the MNI database developers to specify DTI-specific tags.

• Non-rigid Tensor Transformation

There are two issues:

1. What is the best way to do the non-rigid tensor transformation (hard and others are devoting effort on this)?
2. How best to deal with the tensor results (easier)?

MBIRN agreed to take on the second step. A new collaboration between Anders Dale and Mike Miller in which Anders will supply the vector fields for the transformation and tensor fields in native space to Mike Miller. Miller's group will create a tool that will take these inputs give as output the Jacobian matrix and the transformed tensor fields. VETSA data will be used for this pilot project. They will work with Steve Pieper for visualization of the results in Slicer.

• Tractography and Atlases

Jim Fallon will work on constructing another WM atlas. He will start with the best available DTI data (Mori) along with the accompanying structural data. The data will be put into NRRD format for interoperability. He will use the best features of the available programs and work with developers for improvements. The resulting tracts will be collected as an atlas, saving both streamline coordinates and label maps. These results will be provided to the scientific community for download and can be used by algorithmic developers for further research in automating the processing.

Susumu Mori will interact with Steve Pieper and James MacFall to work on the BELL data. They will use LDDMM to transform the T1 data of one set to the other and then use the resulting transformation matrix on the DTI binary atlas data. This should improve the registration.