2009 Summer Project Week Automatic Brain MRI Pipeline
Example Utah segmentation output
Key Investigators
- Utah: Marcel Prastawa
- Iowa: Hans Johnson
- Wash U: Dan Marcus
Objective
Automatic processing of multimodal brain MRI datasets from large populations is a fundamental step for population based analysis. The processing pipeline typically involves registration, segmentation, noise reduction, artifact removal, etc. The Utah registration-segmentation tool is currently being used in Iowa and North Carolina. We plan to further develop the tool to ease integration with Iowa's processing pipeline, and perhaps add new features to the tool.
Approach, plan
We will discuss how the tool is currently used, and find possible customizations that would make it easier to use.
We will design general test cases to simulate how the tool would be used and to ensure consistent stability as we collaborate and exchange code.
Progress
Designed a CLP-based interface for a more flexible connection with other processing steps. This will also allow the loading/saving of program parameters as an XML file (possible future feature in CLP).
Started extending the segmentation parameters to allow testing of bias correction and integration of BRAINSFit.
