If you are referring to the program supplied with MRICron, you might have
the most success asking on the list for that program. In my experience, the
bvecs files generated from dcm2nii always match the output volume.
Therefore things like angulations of the FOV are accounted for and the
gradient directions never need to be flipped or anything like that.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Jesse Brown
Sent: Tuesday, November 03, 2009 5:38 PM
To: [log in to unmask]
Subject: [FSL] dcm2nii bvecs transformations
Does anyone have any information on the *actual* transformations that
dcm2nii is applying to DTI bvecs? From my experience, dcm2nii will
automatically re-orient and swap (in our case Siemens) DICOM files in order
to place them in standard space orientation, i.e. Radiological RL PA IS. It
seems to do this regardless of whether I change the MinReorientMatrix.
Regarding the bvecs, my best guess is that dcm2nii is:
- reading the vector values from the DICOM header (which are in absolute xyz
space), which will be slightly different for each subject when the scan
prescription is oblique
- adjusting the vectors based on the re-orientation from DICOM space (in our
case Neurological RL AP IS) to standard space
If anyone has any information about the operations that dcm2nii is
performing inside the black box, I'd love to hear it. This program seems to
have gained traction along with FSL tools and I'd like to explicitly
understand its operations.
Thanks!
Jesse
|