Usage: fim2 [options] image_files ...
where 'image_files ...' is a sequence of MRI filenames,
options are:
-pcnt # correlation coeff. threshold will be 1 - 0.01 * #
-pcthresh # correlation coeff. threshold will be #
-im1 # index of image file to use as first in time series;
default is 1; previous images are filled with this
image to synchronize with the reference time series
-num # number of images to actually use, if more than this
many are specified on the command line; default is
to use all images
-non this option turns off the default normalization of
the output activation image; the user should provide
a scaling factor via '-coef #', or '1' will be used
-coef # the scaling factor used to convert the activation output
from floats to short ints (if -non is also present)
-ort fname fname = filename of a time series to which the image data
will be orthogonalized before correlations are computed;
any number of -ort options (from 0 on up) may be used
-ideal fname fname = filename of a time series to which the image data
is to be correlated; exactly one such time series is
required; if the -ideal option is not used, then the
first filename after all the options will be used
N.B.: This version of fim2 allows the specification of more than
one ideal time series file. Each one is separately correlated
with the image time series and the one most highly correlated
is selected for each pixel. Multiple ideals are specified
using more than one '-ideal fname' option, or by using the
form '-ideal [ fname1 fname2 ... ]' -- this latter method
allows the use of wildcarded ideal filenames.
The '[' character that indicates the start of a group of
ideals can actually be any ONE of these: [{/%
and the ']' that ends the group can be: ]}/%
[Format of ort and ideal time series files:
ASCII; one number per line;
Same number of lines as images in the time series;
Value over 33333 --> don't use this image in the analysis]
-polref # use polynomials of order 0..# as extra 'orts';
[or -polort #] default is 0 (yielding a constant vector).
Use # = -1 to suppress this feature.
-fimfile fname fname = filename to save activation magnitudes in;
if not given, the last name on the command line will
be used
-corr if present, indicates to write correlation output to
image file 'fimfile.CORR' (next option is better)
-corfile fname fname = filename to save correlation image in;
if not present, and -corr is not present, correlation
image is not saved.
-cnrfile fname fname = filename to save contrast-to-noise image in;
if not present, will not be computed or saved;
CNR is scaled by 100 if images are output as shorts
and is written 'as-is' if output as floats (see -flim).
[CNR is defined here to be alpha/sigma, where
alpha = amplitude of normalized ideal in a pixel
sigma = standard deviation of pixel after removal
of orts and ideal
normalized ideal = ideal scaled so that trough-to-peak
height is one.]
-sigfile fname fname = filename to save standard deviation image in;
the standard deviation is of what is left after the
least squares removal of the -orts, -polrefs, and -ideal.
N.B.: This is always output in the -flim format!
-fitfile fname Image files of the least squares fit coefficients of
all the -ort and -polref time series that
are projected out of the data time series before
the -ideal is fit. The actual filenames will
be fname.01 fname.02 ....
Their order is -orts, then -polrefs, and last -ideal.
N.B.: These are always output in the -flim format!
-subort fname A new timeseries of images is written to disk, with
names of the form 'fname.0001', etc. These images
have the orts and polrefs (but not ideals) subtracted out.
N.B.: These are always output in the -flim format!
-flim If present, write outputs in mrilib 'float' format,
rather than scale+convert to integers
[The 'ftosh' program can later convert to short integers]
-clean if present, then output images won't have the +/- 10000
values forced into their corners for scaling purposes.
-clip if present, output correlations, etc., will be set to
zero in regions of low intensity.
-q if present, indicates 'quiet' operation.
-dfspace[:0] Indicates to use the 'dfspace' filter (a la imreg) to
register the images spatially before filtering.
-regbase fname Indicates to read image in file 'fname' as the base
image for registration. If not given, the first image
in the time series that is used in the correlation
computations will be used. This is also the image
that is used to define 'low intensity' for the -clip option.
