Cornelius,
Unless there is some sort of high frequency artifact, one need not worry
about low-pass filtering. All low-pass filtering would do is
eliminate/attenuate the weighting of certain high frequencies in error
variance estimation but not improve the validity of this estimation if the
noise spectrum is well modeled (as using the AR model in conjunction with
hi-pass filtering will likely effect).
Karl's paper explicates this idea:
1. Friston K, Josephs O, Zarahn E, Holmes A, Rouquette S, Poline J-B. To
smooth or not to smooth? Bias and efficiency in fMRI time-series analysis.
NeuroImage 2000;12(196-208.
Eric
----- Original Message -----
From: "Karsten Specht" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Thursday, June 09, 2005 3:52 PM
Subject: Re: [SPM] How to invoke the low pass filter?
> Dear Cornelius,
>
> > I know that the AR(1) among others are supposed to take care of high
> > frequency noise. But let's suppose I have a block design (SOA 36
> > seconds), and I have the strong suspicion that there is still much (hf)
> > noise in my data - how could I invoke the low-pass filter in SPM2
> > (preferably in batch mode)?
>
> You have to be careful here, since a high frequent noise, having a higher
frequency than your sampling rate of the data, which is very likely, will
turn into a low frequent component in the signal (which is called
'aliasing')
> So, it might be more suitable to adjust the high-pass filter, even it
sounds not intuitive at its first glance.
>
> Good luck,
>
> Karsten
>
> --
> -------------------------------------------------------------
> Karsten Specht, PhD
>
> Department of Biological and Medical Psychology
> & National Competence Centre for functional MRI
> University of Bergen
> Jonas Lies vei 91
> 5009 Bergen
> Norway
> Tel.: +47-555-86279
> Fax: +47-555-89872
> [log in to unmask]
> http://fmri.uib.no/
|