I was sifting through the accumulation of emails from after being away
for a few days, and didn't realise that there had been a few more emails
on the same subject...
Data is lost at the edge of the field of view because resampling interpolates
between the centres of the voxels. It does not extrapolate outside them.
Therefore, a translation of anything over 0.05 voxels the wrong way will
result in the voxel being considered unsampleable. The 0.05 is hard-coded
in spm_vol_utils.c (lines 5 and 550) and also in spm_reslice.m (lines 216
and 372), spm_write_sn.m (line 158) and spm_realign.m (line 277). Changing
the value to something larger (and recompiling the mex files) will preserve
more of the data at the edge of the FOV.
Best regards,
-John
| The point is: we would like to understand where does this problem come
| from, for it's the first time that such a thing happen, and its cause
| remains a mystery to us.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|