Hi FSL list,
I'm seeking some clarification regarding the high pass filter cutoff in the feat GUI. The
feat website states:
"A sensible setting in the case of an rArA or rArBrArB type block design is the (r+A) or
(r+A+r+B) total cycle time. For event-related designs the rule is not so simple, but in
general the cutoff can typically be reduced at least to 50s."
However, I recalled that, during the FSL course, it was suggested to never use a HP filter
less than 100s. I noticed something interesting in this context. If you enter 100s in the
feat GUI, the feat report says
"highpass temporal filtering (Gaussian-weighted least-squares straight line fitting, with
sigma=50.0s)."
Do the values in the FEAT details webpage refer to the values for sigma in the FEAT
report? Should we double the value entered into the feat GUI (e.g., 2 X (r+A+r+B))? Or
is there some other explanation why the feat report lists a value that is half that entered
into the feat GUI?
Thanks,
Greg
|