Hello,
Turns out my data has lesions in both gray matter (probably due to encephalomalacia and/or stroke) and white matter (white matter hyperintensities / WMH).
Firstly, to counter misclassification of white matter as gray matter during fMRI preprocessing in CONN, I'm using LPA on FLAIR (using LST toolbox) to automatically detect WMH and do the lesion filling on ADNI MPRAGE (though this step involves additional coregister:estimate & reslice step to align T1 and FLAIR images in same space). Ultimately I will use this WMH lesion filled MPRAGE for coregistration in my fMRI preprocessing pipeline for resting-state connectivity analysis. I believe doing so will take out the probability of mischaracterisation of white matter as gray matter, hence artifacts arising due to WMH misclassification will be addressed as white matter is later regressed out from connectivity analysis. Please correct me if I'm wrong here.
Secondly using default CONN preprocessing pipeline (but skipping slice-timing correction step due to missing MR acquisition protocol details), I have preprocessed a subject's data with gray matter lesion due to encephalomalacia. It looks like the huge chunk of affected grey matter area has been classified at CSF (observed under Setup.ROIs tab in CONN, please see attachment). Is that how it was supposed to be? Following this, as the affected area classified as CSF will later be regressed out from connectivity analysis, does it mean I don't have to worry about this misclassification like WMH lesions misclassified as GM? Or is there another additional step/analysis that I should look into to address GM atrophy?
Any advise / direction or literature reference will be greatly helpful.
Best,
Dilip
|