[MINC-users] NIfTI to MINC revisited

[Andrew Janke]

Hi Simon,

On 14 August 2012 21:26, Simon Eskildsen <eskild at gmail.com> wrote:
> I found that the mapping to scanner space disappears, while the
> mapping to standard space changes (see the headers below). In this
> case these mappings are identical in the original file.
> The original and recovered image do not align in any viewer I've tried
> (itksnap, mricron, fslview).
>
> I'm no expert in quaternions, direction cosines, and whatnot, so I
> welcome suggestions.

I'll stand by my "sufficient" stance!  I am guessing that these .nii
images come from FSL (by the complete sform and qform
transformations). If so then what you would really need to write is a
fsl2mnc and mnc2fsl where the output of fsl2mnc would be a minc file
and a .xfm. The xfm would represent the transformation from native to
standard space from the FSL nii file (the qform transformation).

The conversion of a fsl .nii file to a minc will (from memory)
concatenate these two transformations into one and when you then
convert back to nii this will only be seen as an sform transformation.
This will manifest itself in what you are seeing.

So you have two options.

1) write fsl2mnc!

2) If you are just doing minc processing (no fiddling with
registration/co-ordinates) then I'd convert back to a .nii .img+.hdr
pair and replace the new header with the original.


a