How To Generate Tracks for IBIS

Here is how you can generate a minc files containing the track information from the clinical GE scanner. To do the same on the research Siemens scanners, go on Ilana’s page: http://www.bic.mni.mcgill.ca/~ilana/diffusion/diffusion_tools.html

Installing the tools

You need to install:

You need to get a copy of GEgrad2mnc.pl from ~laurence/bin

Processing the data

First you need to get the DTI files from the PACS system. You have four different files that you need. The GE scanner uses 25 directions to compute the tensors.

Here is an example of the processing that needs to be done:

  1. Convert dicom to nii using mricron 
  man page: http://www.sph.sc.edu/comd/rorden/mricron/dcm2nii.html
  1. Convert nii to minc using nii2mnc
  2. Write the directions in the minc header using GEgrad2mnc.pl
  3. Register the anatomical with the 4D file using diff_preprocess.pl (only need the -anat option) 
  man page: http://www.bic.mni.mcgill.ca/~ilana/diffusion/diff_preprocess_man.html
  1. Minctensor -mask 
  man page: http://www.bic.mni.mcgill.ca/~ilana/diffusion/minctensor_man.html
  1. Mincfibertrack 
  man page:    http://www.bic.mni.mcgill.ca/~ilana/diffusion/mincFibreTrack_man.html
  *algo FACT
  *namebase 200027277*-mask
  *brute_force
  *mask FA.mnc -gt 0.2
  *roi (In Display create a mask for a seed, usually on the corpus callosum)
  *curvature_constraint_angle 70
  *conn track.mnc (This will be the input for IBIS containing the tracks)

EXAMPLE

cd in the directory with the dicom files/bin/mricron/dcm2nii im* .

gunzip patient.nii.gzip

nii2mnc patient.nii patient.mnc

Luckily this transformation seems to be preserving the starts and direction cosines! :-)

/data/ipl/ipl7/laurence/dti/mincdiffusion/GEgrad2mnc.pl -bvec patient.bvec -bval patient.bval patient.mnc

Before running the diff_process script on the anatomical you have to run the following command on it: minc_modify_header -sinsert zspace:spacing=‘regular__’ global_gado.mnc

The slices that come out of the GE scanner are irregular and nu_correct crashes if you don’t artificially correct for that.

You also need to put your global gado at steps of 1, it does not work with the original 0.5×0.5×1

autocrop -isostep 1 global_gado.mnc global_gado_iso1.mnc

You can only use this following script if the DTI and anatomical were taken during the same scan time. /data/ipl/ipl7/laurence/dti/mincdiffusion/diff_preprocess.pl -anat global_gado_iso1.mnc patient.mnc chiasson_dti.mnc

If the DTI and anatomical are far from each other, other calls to minctracc have to be made.

minctensor.pl -mask global_gado_iso1-n3-bet_mask.mnc patient_dti.mnc

Then open display with the rgb file or the FA file to generate binary seeds for the tracks.

mincFibreTrack -algorithm FACT -namebase global_gado_iso1-n3-bet_mask -brute_force -mask FA.mnc -gt 0.2 -roi seed.mnc -curvature_constraint_angle 70 -conn tracks.mnc

For more details on the algorithm used by mincdiffusion, read: H. Jiang, P. C. van Zijl, J. Kim, G. D. Pearlson, and S. Mori, “Dtistudio: resource program for diffusion tensor computation and fiber bundle tracking,” Comput Methods Programs Biomed, vol. 81, pp. 106–16, Feb 2006.