Adapting the nuisance regression white matter mask to not include brainstem

Submitted by David on

Hi,

are the AllResampled_WhiteMask_09_91x109x91.nii files used for nuisance regression of white matter signal or the FunSpace_ThrdMask_subject_WM.nii files in the segmentation mask folder?I was wondering if it is somehow possible to change the mask that is used for the nuisance regression of white matter before regressing out covariates? In other words, since I'm interested in signals from brainstem I need to make sure that signals from this region are not regressed out as a white matter nuisance covariate. But the white matter mask that is created from the segmentation of SPM normally includes brainstem (specifically, the FunSpace_ThrdMask does but the AllResampled mask does not somehow).

many thanks in advance

 

 

YAN Chao-Gan

Thu, 04/27/2017 - 06:21

Hi, please post a snapshot of your parameter setting (snapshot of "Nuisance Regression Setting" is also needed). It depends on your setting to select the mask.

Hi David,

In your settings, these masks were used:

/Masks/SegmentationMasks/FunSpace_ThrdMask_Sub_001_WM.nii

/Masks/SegmentationMasks/FunSpace_ThrdMask_Sub_001_CSF.nii

Please consider to revise them.

Best,

Chao-Gan

Thanks for the reply,

I'm not sure how to do this, automatically. In other words, I need to remove a specific ROI e.g. Brain-Stem from the Mask BEFORE it will be used in the covariate regression. So this needs to happen right after the masks have been created so it can be used as input to the covariate regression and ideally, this does not interrupt the automatic processing of the data.

This is the tricky part and I'm not sure if this is easily possible with DPARSF, but maybe you know a way to do this?

Hi,

There is not an easy way to do this.

/Masks/SegmentationMasks/FunSpace_ThrdMask_Sub_001_WM.nii

You can edit the mask and continue the remain alaysis starting with Nuisance covarites regression.

Best,

Chao-Gan

David

Mon, 05/22/2017 - 17:04

In reply to by YAN Chao-Gan

Thanks! That is a good idea. The ThrdMasks are not normalized though, so the problem is to somehow remove the normalized ROI (brainstem) from the ThrdMask, which I don't know how to do.  Do you maybe have an idea? :)