Dear Lea,
I believe 18 was picked as the nearest round number that gave something like 95% signal retention at the chosen frequency. fslmaths takes in sigmas, rather than a frequency range, as it allows for filtering with the least assumptions - filters of the given sigma are applied, which can be taken as passing a particular frequency range at a given % retention.
Hope this helps,
Kind Regards
Matthew
> Dear Matthew,
>
> Thank you for the quick reply! How do you arrive at the 1/(18*f*TR)? In particular, where does the 18 come from?
>
> We understand that it's a trade-off between retaining and removing frequencies, but shouldn't there be some convention, especially to make results of fslmaths comparable to SPM and other packages?
>
> Best,
> Lea
|