Core 1 Timeline Notes

From NAMIC Wiki
Revision as of 13:20, 18 December 2006 by Andy (talk | contribs) (Update from Wiki)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Home < Core 1 Timeline Notes

Core 1 Timeline Notes

Notes from the site PIs:

MIT

Following the original timeline, we developed methods for shape-based segmentation, shape analysis and DTI analysis and have started integrating the algorithms into NAMIC toolkit. In addition, we initiated an fMRI analysis theme within NAMIC. Overall, our progress has been according to the original timeline, with a few added milestones and goals for the future development as described below.

Aim 1. Shape based segmentation

We have developed a multi-object hierarchical representations for shape priors in segmentation (years 1-2) and incorporated them into statistical tissue classification (years 2-4). We have validated the method on a limited data set provided by Martha Shenton’s group (Core 3). We are currently working on incorporating the methods into the Slicer and the ITK library and validating the results on larger data sets. Milestones: Integrating shape-based and atlas-driven approaches into a unified framework for multi-object segmentation. Future plans: To apply the segmentation methods to clinical studies in collaboration with Core 3 groups. To fully integrate the methods in the NAMIC toolkit.

Aim 2. Shape analysis

During this past year, we have refined methods for shape statistics by mainly focusing on shape representation and generative statistical models that capture variability of the shapes in a population (years 1-3). We expect to complete the development during next year and to apply the methods to cortical folding patterns and subcortical structures (years 2-5). In addition, we worked on creating software infrastructure in ITK for population studies necessary for shape analysis. This work has been a close collaboration with UNC, GE and Kitware. While not included in the original proposal, this software development project enables representation of population data and its analysis in ITK. Milestones: Defining common infrastructure in ITK for population studies and for shape analysis in particular. Future plans: To finish the development of the novel generative model for shape variability within population. To apply the methods to the clinical studies in collaboration with Core 3 groups. In addition to the original plans, we plan to complete the development of the software infrastructure for the population studies and to integrate of the shape-analysis tools into the pipeline.

Aim 3. Shape and connectivity analysis of DTI data

We have explored several different measures of fiber shape (years 1-3). Moreover, we used theses measures to implement a particular type of a generative statistical model of fiber bundles (years 2-5), creating a white matter atlas. We have been actively collaborating with Core 3 groups to validate the methods and to use them to detect differences in white matter in schizophrenia (years 2-5). The methods for clustering fiber tracks have been fully integrated into the Slicer platform. Milestones: Developing the methods for fiber track clustering and utilizing them in constructing of white matter atlas. Future plans: To apply the methods to clinical data in collaboration with the Core 3 groups.

New development: fMRI analysis

Based on the interests expressed by the Core 3 investigators, we have initiated fMRI-related project within NAMIC. During this past year, we developed a method for brain activation detection that employs Markov Random Fields (MRF) as spatial smoothing priors necessary due to the low signal-to-noise ratio of the fMRI signal. Furthermore, we extended the MRF prior to include anatomical information. The anatomical prior, in the form of a segmented MRI scan, biases the activation detection towards the gray matter and inhibits smoothing of the activation maps across tissue boundaries. We have validated the method on a set of fMRI scans and are currently working on implementation of the detection algorithm in Slicer in collaboration with BWH (Core 2). Our plans include releasing the code into the ITK library to make the method available to a broader community.

UNC

Milestones Achieved

All have been achieved. In some projects we are ahead of schedule in others, we will need the additional time left in year 2 to finalize the results (Aim 1.1). Aim 1.2 should be changes to ‘Develop improved surface shape representations’.

Modifications of Timeline

As we are mostly ahead of schedule this looks fine, but we would like to add a new milestone for the additional research that we are currently doing: Aim 2.4. DTI Atlas Building: begin Middle year 2, end: end of year 4 Also, could you shorten aim 2.2 by one year.


GaTech

We have written Bayesian segmentation and rule-based segmentation code which has been included in ITK. We have statistical/PDE code written some of which is already even in Slicer. Directional based algorithms have C/C++ versions which we want to port into ITK this summer. Finally, we did work out the stochastic evolutions, and have C/C++ code for these as well in the planar case (geodesic active contours). We now are moving to 3D surface evolutions. We have worked closely with Jim Fallon and Steve Potkin at UCI as well as Martha Shenton at Harvard, and have given them code and have worked on their data sets. John Melonakos has been working with the people at Kitware and even has a couple of Insight papers about this work. Delphine Nain is working with Martin Styner of UNC on shape analysis. In short, I think we are right on schedule about what we promised in the proposal.