Hello, 

I am just starting out with DTIPrep through the Slicer extension (version 4.4.0)and have run into the error "Bad DWI: mismatch between gradient image # and gradient vector #!" when I try to use the default protocol. This error has come up on different data sets from two different research centers. Any advice would be greatly appreciated! 

Hayley

Hi Hayley

This is not an error of DTIPrep, but rather when reading your DWI-NRRD file the importer detects an issue, specifically that the number of DWI volumes does not match the number of gradient directions specified in the header. Likely this is a conversion issue when you generated the NRRD file.

1) how did you generate the NRRD file?

2) look at the header (it's ASCII, so you can look at it with any text editor): what is the header information? how many gradients are specified and how many volumes should there be?

Martin

Originally posted by Hayley Clocksin:

Hi Martin, 

I have been importing the DICOM files through Slicer, loading the relevant files and then saving them as a NRRD file.  The header file specifies 25 gradients and 25 volumes. 
Thank you for your help!

Hayley

Hi Hayley
The generic DICOM importer in Slicer does not work well for DWI data. You need to use the DWIConverter module to import your data (see also the QC presentation here (lesson 3 - slide 7) https://www.med.unc.edu/psych/research/niral/download/download-documents/)

The use of DWIConverter in Slicer is a bit clunky, but it's worth using it. On linux you can also use it on the command line:

Slicer --launch DWIConvert -i yourDicomDirectory -o yourOutput.nrrd --useBMatrixGradientDirections

(the last option is not necessary, but provides more accurate b-values and gradient vectors if the BMatrix is stored in the DICOM tags).

Best
Martin

Originally posted by Hayley Clocksin:

Hi Martin,

Thank you so much for your help! It looks like my DICOM files weren't compatible as well, but DWIConverter seems to be working with my NIFTI files. 

Best, 
Hayley