open-discussion
27 Subscribers 
open-discussion > DTIprep on multiband DTI dataset.
Showing 1-2 of 2 posts
Jan 11, 2018 02:01 PM | Tanmay Nath
DTIprep on multiband DTI dataset.
Hi,
I am having issues running DTIPrep on multi-band DTI dataset and it seems DTIPrep cannot handle multiband data. Is it true?
I have a multi-band DTI dataset(MB factor 3) which visually appears to be fine and when I use FSL to compute the motion params, it gives me mean motion of 1.3 and max of 2.9. Ideally, this data set should pass the QC. But when I use DTIPrep to process, the output QCed image has only 8 dirs left out of 137 and the dataset does not pass the QC. Please find attached the screenshot of one of the directions of the QCed image. The image looks distorted too.
Any suggestions? I never had any problem when I ran DTIPrep on the dataset with MB factor 1.
I am having issues running DTIPrep on multi-band DTI dataset and it seems DTIPrep cannot handle multiband data. Is it true?
I have a multi-band DTI dataset(MB factor 3) which visually appears to be fine and when I use FSL to compute the motion params, it gives me mean motion of 1.3 and max of 2.9. Ideally, this data set should pass the QC. But when I use DTIPrep to process, the output QCed image has only 8 dirs left out of 137 and the dataset does not pass the QC. Please find attached the screenshot of one of the directions of the QCed image. The image looks distorted too.
Any suggestions? I never had any problem when I ran DTIPrep on the dataset with MB factor 1.
Jan 11, 2018 03:01 PM | Martin Styner
RE: DTIprep on multiband DTI dataset.
Hi Tanmay
We regularly apply DTIprep to multi band data. There should not be something particular about MB data regarding the QC process that's different from non-MB data.
If the DWI volumes that get rejected by DTIPrep look visually free of intensity artifacts, then your DTIPrep parameters are likely to strict (DTIPrep's default parameter particularly for high b-value images can be too strict). Adjust the parameters of the steps that fails your DWI volume.
Not sure about the distortions, to me it looks like that the voxelsize information in the volume may be incorrect as it seems stretched in inferior-superior direction (or at least incorrect the way AFNI displays it, does it look similarly in fsl if you use fslview, or in Slicer?). It is possible that your acquisition uses a slice overlap and thus that the voxelsize in the slice-direction has been set to the DICOM slice thickness instead of the slice step (difference in slice-position between slices).
Martin
We regularly apply DTIprep to multi band data. There should not be something particular about MB data regarding the QC process that's different from non-MB data.
If the DWI volumes that get rejected by DTIPrep look visually free of intensity artifacts, then your DTIPrep parameters are likely to strict (DTIPrep's default parameter particularly for high b-value images can be too strict). Adjust the parameters of the steps that fails your DWI volume.
Not sure about the distortions, to me it looks like that the voxelsize information in the volume may be incorrect as it seems stretched in inferior-superior direction (or at least incorrect the way AFNI displays it, does it look similarly in fsl if you use fslview, or in Slicer?). It is possible that your acquisition uses a slice overlap and thus that the voxelsize in the slice-direction has been set to the DICOM slice thickness instead of the slice step (difference in slice-position between slices).
Martin