Hi,
I used featquery to obtain mean percent signal-change within a functionally defined ROI. Upon off-line averaging, the group mean percent signal-change for condition A was positive, meaning the group-average percent signal-change estimate was greater than zero (where CopeA was condition A minus Baseline).
I then took the mean/masked timeseries file for this ROI, pulled it into excel and drew up an 8-bin (TR's 1 thru 8, 2 seconds each) peristimulus plot (using a pivot-table). I'm using a slow event-related design, so the 8 TRs following each modeled event are fixation, isolating the event in time. Using TR 1 of 8 as the zero-point anchor, no conditions (including the aforementioned condition A) show signal increases from that zero-point.
I'm trying to digest what the implication is. How could featquery generate a positive percent signal-change estimate for an ROI, yet the timeseries data suggest that all conditions de-activate? Is this something intrinsic to the manner in which percent signal-change is calculated in FSL?
~Jonathan
|