Hi, how to use fslmaths for bandpass filtering, if I want to do 0.01-0.1 bandpass filtering, -bptf should I use 1/(TR * 0.01) and 1/(TR * 0.1), or 1/(2 * TR * 0.01) and 1/(2 * TR * 0.1)?
This is answered in a FSL mailing list post:
There is no perfect formula for converting a bandpass range into sigma - the nature of this filter means there is always a tradeoff between signal retention and filtering. Our preferred formulas are 1/(2\times f\times TR) for the high-pass filter and 1/(18\times f\times TR) for the low-pass filter.
The formulae give sigma in volumes ( taking TR as seconds per volume ). f is the frequency you wish to pass, the “magic numbers” are chosen to (roughly) give 95% signal retention at the chosen frequency.