Mbirn:avi

BIRN AVI Workgroup [1] - January 10th Meeting Minutes

Attendees:

          Timothy Brown <timothy@cis.jhu.edu>
          Polina Golland <polina@csail.mit.edu>
          Jorge Jovicich <jovicich@nmr.mgh.harvard.edu>
          Anthony Kolasny <akolasny@jhu.edu>
          Steve Pieper <pieper@bwh.harvard.edu>

Absent:

          James Fallon <jfallon@uci.edu>
          Bruce Fischl <fischl@nmr.mgh.harvard.edu>
          Mark James <mjames@ncmir.ucsd.edu>

Summary:

1. Reviewed previous minutes and action items. (Kolasny)
   1. All sub aims have been identified for the 6 month and year long goals.
   2. Steve Pieper created a Wiki space for Morph BIRN. The AVI group will be placing their meeting information here.
   3. Vicky Rowley was provided Slicer and LDDMM software versions. Polina mentioned that she should be able to provide her software to Vicky. We are not sure if Polina's software will make the April distribution. Vicky does not have a version of Freesurfer.
   4. birn-avi@nbirn.net has been created. All AVI correspondence will be archived through the BINR-CC list server.
2. Status reports for key areas
   1. segmentation - Fischl

      We have completed a preliminary analysis of the CNR properties of the multi-echo FLASH (MEF) sequence, and have shown that it has superior CNR than MP-RAGE for every tissue class pair investigated (in it's native 16 dimensional space). In addition, MEF has lower distortion that leads to better test-retest repeatability in hippocampal volume measures. Both of these finding will be submitted to HBM2005 as abstracts (as will another, showing it has better test-retest intensity repeatability than MP-RAGE or single echo FLASH). We are now trying to turn the better intrinsic CNR of MEF into better segmentation accuracy.

   2. shape analysis - Kolasny

      • The scaled lddmm processing is complete. This was made possible due to fmri-ucsd allowing our group temporary storage space.
      • 2005 Human Brain Mapping poster will be submitted on Jan. 11th. The JHU statistical analysis group is finishing their analysis.
      • Timothy Brown has a lddmm software development timeline [2]
      • JHU ordered 16TB of additional storage for their BIRN rack. This should allow enough storage for this year project processing.

1. 1. DTI aims - Fallon

      I met with Anders Dale yesterday and we agreed that we need to have another discussion among Anders, Steve Pieper, myself, and the respective techs to decide how much automation we still need to develop the DTI atlas using Slicer. We will have a conference call on this issue this week and then proceed with the atlas development as soon as we agree on whether any other modifications to Slicer are needed to optimally grow the white matter tracts.

   2. visualization - Pieper

      • New 3D Slicer 2.4 version is being released in the next few days.
        • Better cross-platform support on windows, linux, and solaris. Beta version Mac build with essentially identical functionality with release versions. Release candidates are available at http://slicerl.bwh.harvard.edu/snapshots/slicer2.4
        • Release includes LDMMViewer to visualize time-series velocity vectors and deforming hippocampus volumes and models
        • Improved FreeSurfer readers and export functionality supports visualization of .mgh and COR format volumes (grayscale and labelmap) plus cortical surfaces (regular, inflated, spherical) with parcellation and/or thickness overlays.
        • Slicer labelmap editing tools can be used to help clean up automated segmentation results during FreeSurfer processing (requested by Brad Dickerson as clinical collaboration).
        • New nightly build and test infrastructure using Dart -- http://www.na-mic.org/Testing/Slicer
        • Slicer 2.4 is based on VTK 4.4 and ITK 1.8

1. 1. query atlas - Pieper

      • Existing Query Atlas code is being made compatible with current versions of VTK and Slicer (TextureText function required additional debugging).
      • Alex Joyner has been doing additional debugging at UCSD to bring his Query Atlas work forward into Slicer 2.4

1. 1. machine learning - Golland

      At an AVI meeting earlier today Polina was describing her classification tool, which takes as input text datasets (e.g. demographics data, clinical assesment data, subcortical segmentation volumes) and looks for ways of classifying this multidimensional data (e.g. AD vs Controls), gives a visualization tool and calculates statistical significance for the classifiers. Polina is willing to give the source code to the BIRN CC so that they build it into the Portal, distributed on the racks and later add the proper interfaces to the database. She will contact Vicky Rowley for this (vrowley@ucsd.edu). We thought that as a parallel step it would be interesting if Polina could learn how to use the HID to select a few sample datasets, save the CVS file, and use this as input to her classification tool to see if, for example, she picks up significant classifiers that are in agreement to those that one would otherwise be inclined to test via other statistical methods. So my question to you Chris is: could you help Polina access and learn how to use the HID demo pages from UCSD and help her select a group of subjects that would be most interesting for such a study? Probably the best would be to set up a VNC connection between you two and have Chris drive the query assessment tool to select data and show further a few examples of the statistical analyses that can be done.

2. The Next AVI meeting will be at the Morph BIRN meeting in Miami.
3. No new topics to the agenda.

Action Items:

1. Identify people associated with BIRN-DUP project. (Pieper)
2. See if Alex Joyner is still involved with Query Atlas development (Pieper)
3. Include BIRN hippocampus data as an adendum for the Slicer Tutorial (Pieper,Aucoin)
4. Contact Vicky Rowley <vrowley@ucsd.edu> about adding MIT software to BIRN/Rocks distribution. (Golland)
5. Test interoperablility between MIT software and Freesurfer on BIRN/Rocks distribution. (Golland)
6. Contact Christine Fennema <fennema@ucsd.edu> regarding clinical questions for Machine learning (Golland)
7. Make use of the Wiki for AVI dissemination (Kolasny)
8. Provide 2005 Human Brain mapping abstract information to the group. (Kolasny)
9. Provide Vicky Rowley <vrowley@ucsd.edu> a version of Freesurfer for the BIRN portal. (Fischl)

---

Mbirn:Main_Page [3]
Mbirn:Analysis, Visualization, and Interpretation Tools (AVI) [4]