| Table of Contents |
|---|
Where and what
Created in T1 preprocessing folder, for example
Z:\Analysis\Preprocess\ChronicMigraine_7T\T1preprocess\Chronic Migraineurs\Subject_2117\T1_ref_vox0.75_bb_min_-90_-126_-72_max_90_90_108
The brain mask is based on the gray matter (GM) and white matter (WM) files (“c1…”, “c2…” for DARTEL/new segment, “p1…”, “p2…” for CAT). These files are probabilities, and “brain” is classified where GM + WM >= 0.5. In other words, a voxel is “brain” if the sum of probabilities of being gray and white matter is 50% or more.
...
The brainmasklots is the brainmask that has been dilated, which means expanded - puffed out is a good analogy.
...
Identifying problem
If brainmask/brainmasklots look bad, check the c1 & c2 files (or p1 & p2). Here is an example:
...
...
Looking at the gray and white matter files, we see the problem - incorrect gray matter segmentation:
...
When we look at the CSF image (c3), we see the problem - gray matter has been classified as CSF:
...
Possible solution
The most common reason segmentation fails is that the T1 is poor coregistration with the template. In the example about, running “Check coregistration” for “T1 coreg” shows that this subject is not well registered. The first step is then to go to projects, manual coreg, then copy and coreg when done.
...
Problem - unusual brain
In rare cases segmentation fails when a participant has a brain shape that is very different from the template. In that case, it may be that using CAT will help, although this woudl mean running all subjects with CAT.
Otherwise, find the batch job that created the segmentations in the same folder and adjust parameters and run again manually.