Hi Dylan,

I would expect the effect of those small differences in the frequency window of the band-pass filter to be very minor and/or inconsistent across datasets. The specific thresholds used are often a matter of convention, for example the high-pass value 0.008 Hz used by CONN is set to match the default high-pass filter threshold in SPM (which is defined in seconds as defaults.stats.fmri.hpf=128s) but it is not uncommon to see this 0.008 Hz value simply rounded to 0.01 Hz for simplicity, either in the actual computations or when reporting it. Similarly low-pass values of 0.08 Hz or 0.09 Hz are often used to match some of the early papers on functional connectivity which had a Nyquist frequency of 0.0833Hz (from using a TR of 6s), but again this is often just rounded to 0.10 Hz for simplicity. 

Hope this helps

Alfonso

Originally posted by dspets:

Hi all, 

I have seen mixed reports on what bandpass filter to use for resting-state data. Some threads suggest .008-.09 while others suggest .01-.08. Are either of these better than the other? 

Dylan