DWI

DTI PREP

DTIPrep is a dMRI QC software platform developed to identify and correct all common, known dMRI artifacts. The QC process is separated into two phases: (1) a fully automatic phase for quality assessment and artifact correction/removal, and (2) a visual assessment phase for both the DWI volumes and the reconstructed DTI data.

• Go to the dmri directory of a participant. Make sure the bvec and bval files are in the dmri folder.
• In the dmri dir you should create a "dtiprep_nrrd" directory by typing: mkdir dtiprep_nrrd
• while in the dmri dir and type the following command:

[There are currently some issues with converting .nii to .nrrd using this command. convert from dcm instead.]

bsub -q PQ_madlab -I DWIConvert --inputVolume YOUR(.nii)FILE --conversionMode FSLToNrrd --inputBValues YOUR(.bval)FILE --inputBVectors YOUR(.bvec)FILE --outputVolume PARTICIPANT#_dwi.nrrd

• Now a nrrd file will be created. Make sure this .nrrd file goes into the dtiprep_nrrd folder.
• now type the command:

bsub -q PQ_madlab -I DTIPrep

• A GUI will open. Click file --> open DWI and choose your .nrrd file.
• Your dwi brain should show up. Make sure there are no chunks of brain missing.
• Next you need to do the QC. Click on the protocol tab, click default, insert all your scan parameters, and save the protocol somwhere in the participant folder. Load the protocol and click "Run By Protocol". This should take ~30 minutes.
• After this is done check the QCed filed in the .nrrd file.

More information about DTIprep: https://www.frontiersin.org/articles/10.3389/fninf.2014.00004/full

the automatic QC phase performs the following steps:

1. DICOM to NRRD conversion

2. Image information checks (ensuring correct image dimensions, spacing, and orientation).

3. Diffusion information checks (ensuring correct diffusion gradient orientations, gradient b-values).

4. Rician noise removal on raw DWI volumes.

5. Inter-slice brightness artifact detection via normalized correlation analysis between successive slices within a single DWI volume.

6. Interlaced correlation analysis for detection and removal of “venetian blind” artifacts and motion within a single DWI volume.

7. Co-registration to an iterative average over all the baseline images.

8. Eddy-current and motion artifact correction, including appropriate gradient direction adjustments.

9. Residual motion detection to ensure all DWI volumes are well registered.

10. Reconstruction of the DTI data and computation of DTI property maps. Currently, DTIPrep implements a single tensor model via weighted least-squares fit (Goodlett et al., 2007) followed by computing the standard tensor property maps, which include fractional anisotropy (FA) , mean diffusivity (MD) , axial diffusivity (AD) and radial diffusivity (RD).

11. Directional artifact detection/correction. This step is a recent algorithmic development and will be discussed in more detail below in section 2.1.

DWI Convert

In the next step we convert our .nrrd files to .nii files (fsl files). For the SEQTRD task, you would need to run this command after going to the following directory:

/home/data/madlab/data/mri/seqtrd/7831##/dmri

Where the number signs above will be different for each different participant.

The command is the following:

DWIConvert --conversionMode NrrdToFSL --inputVolume dtiprep_nrrd/783125_dwi_QCed.nrrd --outputVolume 783125_dwi_QCed.nii.gz --inputDicomDirectory . --outputBValues bval_QCed --outputBVectors bvec_QCed --outputDirectory .

This will convert the nrrd files to fsl files. The file you will convert is the QCed.nrrd file or the QCed_VC.nrrd file depending on whether or not you did any censoring during the visual quality control step. If you didn’t just use the QCed.nrrd file (like the example above).

This tells the software to create an output volume called 783125_dwi_QCed.nii.gz in the directory that you will be running this command in (e.g., /home/data/madlab/data/mri/seqtrd/7831##/dmri). It will output Bvalues and Bvectors that you should call bval_QCed and bvec_QCed respectively. The period following --outputDirectory tells the software to save everything to the current directory. The —inputDicomDirectory period is residual and doesn’t do anything but needs to be included in your script.

BedPost X

Probabilistic Tractography

K-means Clustering