 |
 |
Hi Thomas, Well it seems like your 100 is much too high, given your low frequency blocks. 128s should have somewhat less effect on your task frequency signals - hence I guess the change in activations - but still seems to high. Are your blocks randomised? If not, as the FEAT advice suggests, 4*56s is appropriate. Eugene On 17 July 2012 09:43, Goldschmidt, T. <[log in to unmask]> wrote: > Dear FSL-experts,**** > > ** ** > > I am struggling with a highpass filter question in FEAT. I read in the FSL > guide that the filtering shouldn’t be shorter than the maximum stimulation > time. In my case, I guess, 4 x 56 secs (see below). However, in the > literature, in similar tasks, high pass filtering of 128 secs is used. And > what is striking to me, is that there are huge differences in the > activation patterns when different highpass filters are used. Is there > someone who could give me an advice which filtering to use? **** > > I would be very glad about some insights in this (hopefully clear :) > question. **** > > ** ** > > Cheers,**** > > ** ** > > Thomas**** > ------------------------------ > > *From:* Goldschmidt, T. > *Sent:* maandag 16 juli 2012 11:37 > *To:* [log in to unmask] > *Subject:* highpass filter in block design**** > > ** ** > > Hi FSL-experts,**** > > ** ** > > I am starting up an experiment with an n-back-block design. There are 4 > conditions (each 56 sec long). The first condition (n=0, let’s call it “a”) > is not really a working memory task but more an attentional task. Then > there are the three working memory conditions (n=1, n=2 and n=3, let’s call > them A, B and C respectively) with increasing working memory load. So, in > total my design looks like aBCD. I wonder now how long I should choose the > highpass filtering. Should it be longer than 4x56 sec or longer than 3x56 > sec and how much longer? When I compare different highpass filters (f.ex. > 100 – the standard value - and 128sec – often done in the literature) there > are huge differences in the activation pattern.**** > > ** ** > > I would be glad to get some advice!**** > > ** ** > > Best**** > > ** ** > > Thomas **** > > ** ** > > ** ** >