Hi Jeff,
> BTW - we were using a resize_img script, that appeared to be based upon
> J.Ashburner script functions
Just to note, if this is my version of this script,
http://www.cs.ucl.ac.uk/staff/gridgway/vbm/resize_img.m
I do admit in the code that it's ripped off from JA ;-)
> but the resulting files were reduced in image intensity
I'm a bit puzzled how that could happen... could you give more
information about what you observe (e.g. before/after intensity at a
voxel and/or mean intensity over image)
> and appeared smoothed
As Volkmar says, some loss of accuracy is unavoidable with
reslicing/interpolating images. My resize_img just uses linear
interpolation (see line 106), which is not ideal. You'll get less
blurring (but possibly some slight ringing artefacts) if you change
this to use sinc interpolation -- see the help for spm_slice_vol.
Better still is to use B-spline interpolation, as used in spm_reslice.
As Volkmar says, using spm_reslice (realign/reslice or whatever it's
called in the SPM2 GUI) requires that you have a reference image
available. One option would be to use resize_img to get a blurry but
hopefully geometrically correct reference image, and then to use this
reference with spm_reslice. [I/someone should probably change
resize_img to use B-spline interp, or possibly better still, patch
spm_reslice to allow use without a reference image...]
Note that imcalc just uses spm_slice_vol as resize_img does, so it
won't give better results.
Best,
Ged.
|