Mbirn: Updates-4

Feb 24, 2005 (J. Sacks)

Revised summary:

• As part of the fBIRN a new tool for multi-volume preview and processing , called Ibrowser , is being developed for 3D Slicer by Wendy Plesniak. A key feature of Ibrowser will be the ability to load a multi-volume set as a time series. Such time series can then be analyzed using various tools. The Ibrowser can currently load and display multi-volume sequences of images in several data formats in Slicer. Ibrowser will be extended to load perfusion sequences, enable their processing through its GUI and permit animated preview and saving. Early work on these features is already underway. Once a sequence has been motion corrected by registration, other existing Slicer tools for defining regions or volumes of interest and for performing various measurements and computations on volumes can be applied to the data. We may discover that additional specialized measurement tools may also be required to do the necessary Investigation analysis.

• A critical step for being able to perform the myocardial blood perfusion analyses is firstly to co-register the set of volumes in the time series. Being able to do this quickly and accurately is a non-trivial task. The immediate goal for the group working on this project is to decide on the best method for performing the co-registration, and then to implement it within the Ibrowser module.

1. Gunner Farneback, under the guidance of C.F. Westin, has been working on a non-rigid registration algorithm which is being tested. Presumably it could be used standalone. The algorithm uses both the ITK and MATLAB packages. It does give good results compared with the current method. However, since it is tied to MATLAB, it cannot be incorporated in 3D Slicer’s Ibrowser , which will be using only ITK.
2. Steve Pieper and Wendy Plesniak have proposed that as an initial step, a rigid co-registration be performed manually in Ibrowser on key frames to be interpolated across intermediate frames. Work on this is starting up.
3. In the long term it will be necessary to develop new co-registration tools for multi-volume time series in Ibrowser. It is hoped that techniques being successfully used in neuroscience registration will be transferable and will produce far greater accuracy, possibly down to the the voxel level

Start Date: January, 2005
Completion Date: no later than December,2005

The criteria for success of this project are the completion of the following:
Deliverables:

• (1) Comparison of available registration methods (method currently used by Adler/Kwong, Gunnar Farneback’s method, Pieper-Plesniak proposal).
  Status: currently in process Expected completion date: March 31, 2004
• (2) Longer term development of multi-volume registration methods to be used in Ibrowser
  Status: initial work on Ibrowser has been completed by Wendy Posniak
• (3) Incorporation or development of tools in Slicer for the quantitative analysis of myocardial blood perfusion.
• (4)Application to the BWH/GCRC myocardial blood perfusion study (may begin after (1) is completed)

Milestones for GCRC meeting in April:

• (a) Results of investigation into which registration methods should to be used in Ibrowser
• (b) A more refined definition of the project
• (c) A revised workplan

---