Hi - sigma (in seconds) is FWHM (in seconds) / 2.35
Cheers



On 21 Oct 2011, at 07:47, Richard Bethlehem wrote:

Dear FSL users,

I have a quick question about band-pass filtering with fslmaths. I have already preprocessed all of my resting-state fMRI data (slice-time, reallign, coreg, norm & smooth) and I want to apply a high and low-pass filter to it as well, before running analysis on it. I haven't used the fslmaths -bptf before so I am a little unsure about the calculation of the lp and hp sigma values. I thought it had to be be calculated as follows: sigma = (1/hz)/TR

And then the code would be:

"fslmaths input -bptf [hp_sigma] [lp_sigma] output"

So if I want to filter my data 0.08<f<0.009 then the calculation would be (given a TR of 1.3):
hp_sigma = (1/0.009)/1.3
lp_sigma = (1/0.08)/1.3

Now I a little confused after reading some posts related to this on the fsl archive here, where some people also seems to divide this by a factor 2.35. Which is the correct formula for the sigma values: (1/hz)/TR   or  ((1/hz)/2.35)/TR?

Hope one of you can give me a difinitive answer on this! Thanks in advance.

Best,

Richard



---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------