Hello SPM experts,
I want to estimate the connectivity between two VOIs and use the VOI batch to extract the time-series from the first-level model. The first-level model has covariates for task, movement, WM, CSF and a global signal. The global signal seems to explain a large share of the variance. Hence, the complete model consists of 11 covariates plus the constant.
I want to extract the time-course after adjusting for the variance that is explained by all the covariates together. First I thought I needed to set up the f-contrast as eye(11) but the resulting time-series is almost perfectly correlated with the non-adjusted time-series so that can't be it, right? I tried to only adjust for the global signal (third regressor in design matrix, f-contrast: 0 0 1), which resulted in a time-series that is much weaker correlated (0.6) with the non-adjusted time series. Hence, I figured if I wanted to eliminate the total variance that is explained by the covariates of no interest I have to set up a f-contrast like this: 1 1 1 1 1 1 1 1 1 1 1 0. The results seem quite plausible but I really want to make sure that I'm doing everything right here. I would highly appreciate if anybody could help me out here! Thanks!
Peter
|