Any filtering performed outside SPM will compromise the statistic to some extent, unless SPM is informed about the degrees of freedom used by the filter. There are several papers on filtering fMRI data, in case you are interested, most of my viewpoints on this subject is included in a NeuroImage paper from January 2006. "Non-White Noise in fMRI: Does modelling have an impact?

Torben Ellegaard Lund

Assistant Professor, PhD

The Danish National Research Foundation's Center for Functionally Integrative Neuroscience (CFIN)

Aarhus University

Aarhus University Hospital

Building 30

Noerrebrogade

8000 Aarhus C

Denmark

Phone: +4589494380

Fax: +4589494400

http://www.cfin.au.dk

[log in to unmask]

Den 10/08/2007 kl. 18.25 skrev Ryan Miller:

Hi guys,
Please ignore the bulk of my last email entitled "filtering fMRI data", I just realized upon sending it that the butter filter specified is actually a low-pass filter. However, I would like to ask the question of whether or not there is preference to do low-pass filtering before modeling (as with a butter filter), or rather to account for serial autocorrelations during modeling (which in my limited understanding seems to accomplish a similar job to low-pass filtering). Thanks again,

Ryan