3dTfilter takes as input a dataset, filters the time series in
each voxel as ordered by the user, and outputs a new dataset.
The data in each voxel is processed separately.
The user (you?) specifies the filter functions to apply.
They are applied in the order given on the command line:
-filter rank -filter adaptive:7
means to do the following operations
(1) turn the data into ranks
(2) apply the adaptive mean filter to the ranks
** This program is a work in progress, and more capabilities
will be added as time allows, as the need arises, and as
the author's whims bubble to the surface of his febrile brain.
** This program is for people who have Sisu.
At least one '-filter' option is required!
The FunctionName values that you can give are:
rank = smallest value is replaced by 0,
next smallest value by 1, and so forth.
** This filter is pretty useless.
adaptive:H = adaptive mean filter with half-width of
'H' time points (H > 0).
** At most one 'adaptive' filter can be used!
** The filter 'footprint' is 2*H+1 points.
** This filter does local smoothing over the
'footprint', with values far away from
the local median being weighted less.
adetrend:H = apply adaptive mean filter with half-width
of 'H' time points to get a local baseline,
then subtract this baseline from the actual
data, to provide an adaptive detrending.
** At most one 'adaptive' OR 'adetrend' filter
can be used.
despike = apply the 'NEW25' despiking algorithm, as in
despike:H = apply the despiking algorithm over a window
of half-with 'H' time points (667 > H > 3).
** H=12 is the same as 'despike'.
** At most one 'despike' filter can be used.
detrend:P = (least squares) detrend with polynomials of up
order 'P' for P=0, 1, 2, ....
** At most one 'detrend' filter can be used!
** You can use both '-adetrend' and '-detrend',
but I don't know why you would try this.
3dTfilter -input fred.nii -prefix fred.af.nii -filter adaptive:7
Author: The Programmer with No Name