NA-MIC/Projects/External Collaboration/Measuring Alcohol and Stress Interaction

Return to Project Week Main Page

Key Investigators

• BWH: Kilian Pohl
• Virginia Tech: Chris Wyatt, Vidya Rajagopalan
• Kitware: Luis Ibanez

Objective

The objectives of this project are to:

• Optimize the EM Brain Classifier in Slicer on MRI images of vervets.
• Segment data from N subjects.
• Optimize GM/WM/CSF segmentation so that calculation of GM/WM volume and ratios between the two states can be automated.
• Optimize hippocampus, putamen and caudate segmentation so that volume comparison between states can be performed.

Approach, Plan

We will use the current implementation of the EM Segmenter in Slicer. Using a manual segmentation as an estimate of ground truth in a single subject, we will optimize the EM segmenter parameters over the classification error and running time.

Progress

Prior to the project week:

• Study data on 8 subjects at two time points (alcohol naive and post-induction) organized
• GM/WM/CSF atlas will be constructed for the study data.

During Project Week

• Optimized segmentation parameters for vervet data
• Improved accuracy of subcortical segmentation in rhesus data set

Segmentation of major tissue classes in for vervet subject

Subcortical Segmentation of Rhesus subject

• Developed an ITK utility for multiresolution demons registration. (Source code available for download from the NAMIC sandbox)

Checker Image of source and target before registration

Checker Image of source and target after registration