search

canlab / CanlabCore

Core tools required for running Canlab Matlab toolboxes. The heart of this toolbox is object-oriented tools that enable interactive analysis of neuroimaging data and simple scripts using high-level commands tailored to neuroimaging analysis.
https://canlab.github.io/CanlabCore/
Other
146 stars 80 forks source link

concatenation of runs and regressors #18

Closed inej closed 7 years ago

inej commented 8 years ago

Hi,

A quick question about design matrix: when running the GLM, is it better to concatenate the data (various runs), the movement regressors, the wm and csf avg. signal in the matrix?

yonestar commented 8 years ago

Hi,

I think it may be better not to concatenate the nuisance regressors across runs. For example, its possible that effects of head motion will differ between runs at a given voxel, I think.

We often concatenate the regressors of interest across runs, though it should not matter much whether you do that or not.

Best Yoni

On Mon, Aug 22, 2016 at 1:56 PM, inej notifications@github.com wrote:

Hi,

A quick question about design matrix: when running the GLM, is it better to concatenate the data (various runs), the movement regressors, the wm and csf avg. signal in the matrix?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/canlab/CanlabCore/issues/18, or mute the thread https://github.com/notifications/unsubscribe-auth/AETgjhCGwzRJIAva-DOiDp8p_sENCDlqks5qif7rgaJpZM4JqPZz .

danweflen commented 7 years ago

Has this problem been solved? Or did we decide that it was better not to modify the code?

sgeuter commented 7 years ago

if you concatenate regressors across runs, you will to manually adjust the highpass filtering and AR model. that would be something to manually after SPM design specification and before design estimation.