Hi Mike,
the intent was to be independent of the prescription of the slices so that we could always depend on voxel coordinates having some anatomical meaning, regardless of what crazy slice orientation was prescribed. Not sure if Doug has some easier work around for you to prevent the rotation.
cheers Bruce
On Thu, 16 Jun 2011, Michael Harms wrote:
Hi guys,
I'm curious why the conform step of mri_convert (in generating mri/orig.mgz) automatically interpolates oblique acquisitions. Is this a historical legacy? Something mandated by other elements of the FS pipeline? As a consequence of this behavior, if you have an oblique acquisition that was intentionally aligned to the actual anatomy of a given subject's brain (e.g., using AutoAlign to generate a AC-PC aligned acquisition), unless one manually alters the qform in the input NIFTI file to remove the "oblique" components of the xform, there will be an interpolation in the generation of the mri/orig.mgz, which brings with it blurring (i.e., the same sort of concerns that prompt one to use a single MPRAGE, rather than the average of two MPRAGEs)
Just curious...
thanks, -MH