 |
 |
Dear David, I tracked down your message on the FSL website (it somehow got into an attachment that was removed in the mailing so it was just blank for us all!) The current implementation of cutoffcalc is known to be too conservative sometimes and it looks like this is the case for you here. If that occurs then just ignore the value and use your knowledge of the design to choose a value, but double-check that it does not alter the EVs noticeably, as if it does (compared to no filtering, or a very high cutoff filter) then you are losing substantial signal from your design and that will lead to low statistical power. Don't worry if it just changes at the ends though - that is to be expected and not a particular problem. Hope this helps. All the best, Mark On 3 Feb 2011, at 15:18, David Soto wrote: > > Hi, > > > following the contents of a recent posts I've used the cutoffcalc command to > stimate the most adequate high pass filter for my design, > after saving my first level design matrix without specifying a highpass filter > (by entering 9999 following Mark Woolrich suggestion) and then doing > > cutoffcalc --tr=2.2 -i cutofffl.mat > > the result was 827... is not that too much? I wonder what would this filter be doing... > > my paradigm is an event related design, and there 350 volumes per run, tr=2.2 > > design matrix used is attached........ > > thanks, > > david > >