questions > Files generated when DTI sequence converted
Showing 1-2 of 2 posts
Jun 27, 2016 05:06 PM | Brian Biekman
Files generated when DTI sequence converted
To Whom It May Concern:
When I drag and drop a DTI sequence from a Philips scanner into the GUI with the option "Compressed FSL (4D NIFTI nii)", quite a few output files are generated: the expected .nii.gz file, an .hdr and .img file. Then there are files with the same name except with an "x" in front and the same files except with an "x" in front and an "A" at the end. However, bval and bvec files are only generated for the files with the "x" in front and "x" in front and "A" at the end. What do these represent? And are any of these .nii files flipped in orientation from radiology to neurology convention? I have attached a screenshot. Thank you.
-Brian Biekman
When I drag and drop a DTI sequence from a Philips scanner into the GUI with the option "Compressed FSL (4D NIFTI nii)", quite a few output files are generated: the expected .nii.gz file, an .hdr and .img file. Then there are files with the same name except with an "x" in front and the same files except with an "x" in front and an "A" at the end. However, bval and bvec files are only generated for the files with the "x" in front and "x" in front and "A" at the end. What do these represent? And are any of these .nii files flipped in orientation from radiology to neurology convention? I have attached a screenshot. Thank you.
-Brian Biekman
Jun 28, 2016 01:06 PM | Chris Rorden
RE: Files generated when DTI sequence converted
It might help to know a bit about the software you are using
(dcm2nii which comes with MRIcron vs dcm2niix which comes with
MRIcroGL) and the operating system you are using (Windows, OSX,
Linux). I presume from the screenshot you are using Windows. In
general, unless you are analyzing legacy data I would suggest you
use dcm2niix/MRIcroGL. The older tools are considered mature and
stable but are not actively extended. In terms of development
MRIcro was 1999-2004, MRIcron was 2005-2009 (MRIcro for NIfTI),
MRIcroGL is 2010-present.
1.) Philips files where derived ADC maps are included are not suitable for analysis with FSL or other tools, so the 'x' files exclude these. There is no sensible bval/bvec for an ADC map so the raw 4D volumes that include these do not include these text files.
https://www.nitrc.org/forum/forum.php?thread_id=6243&forum_id=4703
2.) Image rotation transforms are stored in the S-Form and Q-from of the header. Therefore, regardless of the order the vendor saves the image information to the disk, any NIfTI-aware software can correctly report orientation and (if desired) display images oriented in a specific orientation. However, viewing software that ignores the spatial transform of the header is at the mercy of the vendor's convention. Note that with 3D acquisitions dcm2niix will losslessly reorient images so that the NIfTI spatial transformation matrix roughly matches the identity matrix, providing a consistent alignment (with residual differences stored in the S and Q Form). However, this is not done for 2D EPI sequences (DTI, fMRI) as it would confound current slice timing tools.
1.) Philips files where derived ADC maps are included are not suitable for analysis with FSL or other tools, so the 'x' files exclude these. There is no sensible bval/bvec for an ADC map so the raw 4D volumes that include these do not include these text files.
https://www.nitrc.org/forum/forum.php?thread_id=6243&forum_id=4703
2.) Image rotation transforms are stored in the S-Form and Q-from of the header. Therefore, regardless of the order the vendor saves the image information to the disk, any NIfTI-aware software can correctly report orientation and (if desired) display images oriented in a specific orientation. However, viewing software that ignores the spatial transform of the header is at the mercy of the vendor's convention. Note that with 3D acquisitions dcm2niix will losslessly reorient images so that the NIfTI spatial transformation matrix roughly matches the identity matrix, providing a consistent alignment (with residual differences stored in the S and Q Form). However, this is not done for 2D EPI sequences (DTI, fMRI) as it would confound current slice timing tools.