This is an old revision of the document!
lncd MNI 2018 (grid size mismatch)
An fsl bug created 2mm and 2.3mm resampled MNI temples with an extra row of zero voxels in the image. (original 1mm was unmodified/unaffected). This off-by-one template was used for cog and PET Study. The correct MNI template is used for preprocessing 7T Magnetic Resonance Imaging and onward.
LNCD Tools's fixto1809c
runs flirt to resample. But 3dresample
might be sufficient (20240418WF) – the origin information within the nifti header is still correct: entering a x,y,z MNI coordinate in afni will jump to the same place in any of the template underlays – but with normal issues of a discrete grid.
The matrix i
size for 2mm should be 97 but is 96 in 2mm, and for 2.3mm should be 83 but is 100. Similarly the left edge of the x
coordinate in the image should be -96
but is -95
in 2mm and -95.45
in 2.3mm.
# 1mm -- actual MNI 2009c 3dinfo -n4 -extent /opt/ni_tools/standard/mni_icbm152_nlin_asym_09c/mni_icbm152_t1_tal_nlin_asym_09c.nii 193 229 193 1 -96.0 96.0 -96.0 132.0 -78.0 114.0 # 2mm 3dinfo -n4 -extent /opt/ni_tools/standard{,_old}/mni_icbm152_nlin_asym_09c/mni_icbm152_t1_tal_nlin_asym_09c_2mm.nii |perl -slane 'print join(" ", map {$_=sprintf "%.0f", $_} @F)' 97 115 97 1 -96 96 -96 132 -78 114 # correct 96 114 96 1 -95 95 -95 131 -77 113 # lcnd off by one # 2.3mm 3dinfo -n4 -extent /opt/ni_tools/standard{,_old}/mni_icbm152_nlin_asym_09c/mni_icbm152_t1_tal_nlin_asym_09c_2.3mm.nii |perl -slane 'print join(" ", map {$_=/\./?sprintf("%.2f", $_):$_} @F)'|column -t 83 99 83 1 -96.00 92.60 -93.40 132.00 -78.00 110.60 84 100 84 1 -95.45 95.45 -95.85 131.85 -77.45 113.45
In afni, you can see the viewer move the crosshair (coordinate says the same) and mask when the underlay is changed.
And how it effects placement