I do not know if an adjustment for an F-contrast is possible with
VOI-extraction in Marsbar. However, I strongly recommend to remove all
unwanted variance from the time-series which you can explain by
adjusting your data according to a suitable F-contrast. I do not know
exactly the reason why the raw time-series has this U-shape, but it
seems to be quite common for un-adjusted data. Either the adjustment
according to an F-contrast or the high-pass filtering of the data should
change the shape of your time-series.
On Sun, 2013-12-15 at 17:11 +0000, Mark wrote:
> For PPI analysis in spm8, we have to input the time series of a seed region. I know spm8 itself can extract time series (by pressing "eigenvariate"). However, in all my subjects, I already have their time series data in a ROI extracted by MarsBar (select SPM.mat --> "Data" --> "Extract ROI data (default)" --> Select my ROI mask --> "Data" --> Export the time series data), which were saved as VOI-ROI.mat file (it seems that files containing time series that can be selected as input in PPI must have a name initiating with "VOI").
> Can I use this MarsBar-produced time series (i.e. VOI-ROI.mat) as VOI input in PPI batch module in spm8? Or, the time series must be "eigenvariate" data? (However, eigenvariate time series of my fmri data are in U-shape, with an extremely large the very first and very last number).