From NITRC Wiki

INDI Preprocessed DTI Data - Beijing Enhanced Sample

This dataset allows to perform probabilistic tractography with BedpostX (FSL, FMRIB) or standard-space analysis of DTI scalars on the Beijing Enhanced neuroimaging sample.

Original summary of the Beijing Enhanced Sample: These data include 180 healthy controls from a community (student) sample at Beijing Normal University in China. Compared to the Beijing_Zang dataset in the 1000 Functional Connectomes Classic collection these data include IQ scores for a subset of participants (n=55) and a 64 directions DTI scan for all participants.

INDI DTI Data preprocessing was performed by András Jakab, MD. The processing pipeline used the binaries from the FSL software package by the FMRIB workgroup, Oxford. If you use this data in a publication please make sure you acknowledge, András Jakab (University of Debrecen, Hungary and Computer Vision Laboratory ETH Zürich Switzerland), Yufeng Zang (Beijing Normal University, State Key Laboratory of Cognitive Neuroscience and Learning), the International Neuroimaging Data-sharing Initiative (INDI), and the The Neuro Bureau.

Please note that the IDs of subjects have been changed to a 3-digit ID (range 001-180) in order to ease scripting and mass-processing.

e.g.: 1060223 -> 001

The conversion between the legacy ID and the new IDs are included in the CSV dataset (incl. in each tarball). The phenotypic data (age, gender, IQs) of the original .csv dataset have also been added.

We have setup the indi-preproc-dti forum for users to post questions and discuss issues related to the preprocessing. In the interest of making this data available as quickly as possible, we have not performed any quality control on the raw or preprocessed data. We are hoping that the community will help with this process, please post your quality control judgements to the user list so that others may benefit from your efforts.

Downloads

DTI Prepocessed data can be downloaded from here.

Description of included files

For detailed description of the image acquisition protocol, see the original INDI website: fcon_1000.projects.nitrc.org/indi/retro/BeijingEnhanced.html

Each tarball contains numbered directories (for subject ID disambiguation, see introduction). All image files are in gzip compressed NIFTI format (.nii.gz). A phenotypic descriptor (.csv) of the entire subject population (n=180) can be found in each archive.

Each 3-digit directory contains the following subdirectories:

/..

/bedpostx.bedpostX/   -   (BedpostX results directory for performing probabilistic tractography)

bvals (b-values for each diffusion direction)
bvecs (b vectors of the diffusion weighting)

... bedpostx result files (dyadsN,meand,meanfN,meanphN ...) for detailed description on what BedpostX does and what these files represent, see www.fmrib.ox.ac.uk/fsl/fdt/fdt_bedpostx.html

nodif_brain_mask.nii.gz (the brain mask of the diffusion images, created by using bet on B0 images)

/scalars/  -  (DTI scalars, diffusion space)

DTI_FA.nii.gz  (fractional anisotropy image, diffusion space)
DTI_L1.nii.gz  (1st eigenvalue image, diffusion space)
DTI_L2.nii.gz  (2nd eigenvalue image, diffusion space)
DTI_L3.nii.gz  (3rd eigenvalue image, diffusion space)
DTI_MD.nii.gz  (Mean Diffusivity image, diffusion space)
DTI_MO.nii.gz  (Mode image, diffusion space)
DTI_S0.nii.gz  (B0 image - no diffusion weighting - diffusion space)
DTI_V1.nii.gz  (Principal eigenvector image, diffusion space)
DTI_V2.nii.gz  (2nd eigenvector image, diffusion space)
DTI_V3.nii.gz  (3rd eigenvector image, diffusion space)

/scalars_standard/    -   (DTI scalars, non-linearly transformed to standard MNI152 space.)

DTI_FA.nii.gz (fractional anisotropy image, MNI152 reference space)
DTI_L1.nii.gz (1st eigenvalue image, MNI152 reference space)
DTI_L2.nii.gz (2nd eigenvalue image, MNI152 reference space)
DTI_L3.nii.gz (3rd eigenvalue image, MNI152 reference space)
DTI_MD.nii.gz (Mean Diffusivity image, MNI152 reference space)
DTI_MO.nii.gz (Mode image, MNI152 reference space)
DTI_S0.nii.gz (B0 image - no diffusion weighting - MNI152 reference space)
DTI_V1.nii.gz (Principal eigenvector image, MNI152 reference space - transformed reoriented with a nonlinear deformation field using vecreg, FSL)

/transforms/     (Transformation files)

anat2ref-warp.nii.gz (anatomical MPRAGE T1 image to MNI152 standard space, nonlinear warp field, FNIRT)
''anat2ref.mat(anatomical MPRAGE T1 image to MNI152 standard space, linear affine transformation, FLIRT)
diff2ref-warp.nii.gz(diffusion space to MNI152 standard space, nonlinear warp field, FNIRT)
diff2ref.mat(diffusion space to MNI152 standard space, linear affine transformation, FLIRT)
ref2diff-warp.nii.gz

Details of the DTI Processing Pipeline

DTI preprocessing included the use of BASH scripts and FSL binaries (FSL, FMRIB, Oxford).

1. Pre-processing
   1. Conversion of DTI DICOM files to data.nii.gz, bvals, bvecs: using DCM2NII from MRICRON.
   2. Brain extraction from "no diffusion - B0" images using the following command:

      bet $dir/data $dir/nodif_brain -m -f 0.4 -R

2. Calculation of DTI Scalars
   1. Calculation of scalar data using FSL: 

      dtifit -k $dir/data

3. BedpostX - Modeling of local diffusion parameters - Bayesian Estimation of Diffusion Parameters Obtained using Sampling Techniques
   
   1. Bedpostx run with 2 fiber populations modeled:

      bedpostx $dir -n 2

4. Standardization
   1. Linear registration of diffusion space into standard space. The transformation was determined that registers the FA image to the FMRIB58 FA template.

      flirt -ref ${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain -in $dir/DTI_FA -omat $dir/diff2ref.mat

   2. Refine linear registration with non-linear registration.

      fnirt --in=$dir/DTI_FA --aff=$dir/diff2ref.mat --cout=$dir/diff2ref-warp.nii.gz --config=FA_2_FMRIB58_1mm

   3. Transformation of DTI scalars and nodiff_brain image to standard space.

      applywarp --ref=${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain --in=DTI_FA --warp=diff2ref-warp.nii.gz --out=$dir/scalars_standard/DTI_FA.nii.gz

   4. Transformation of eigenvector images to standard space.

      vecreg -i $dir/DTI_V1.nii.gz -o $dir/scalars_standard/DTI_V1.nii.gz -r ${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain -t $dir/diff2ref-warp.nii.gz

   5. Determination of the reference space to diffusion space nonlinear deformation.

      invwarp -w $dir/diff2ref-warp.nii.gz -r $dir/nodiff_brain -o $dir/ref2diff-warp.nii.gz

   6. Standardization of MPRAGE T1 Image and saving ref2anat and anat2ref deformation fields. For more details, visit FSL's FNIRT website.

Quality Metrics

Any feedback on the quality of pre-processing is appreciated (e.g. quality of non-linear transformations, misregistrations).

Please post your feedback to indi-preproc-dti or send an email to jakab@vision.ee.ethz.ch

References

1. Yan, C., Gong, G., Wang, J., Wang, D., Liu, D., Zhu, C., Chen, Z.J., Evans, A., Zang, Y., He, Y., in press. Sex- and Brain Size-Related Small-World Structural Cortical Networks in Young Adults: A DTI Tractography Study. Cerebral Cortex.