Skip to content

0A. FreeSurfer reconstruction

Fa-Hsuan Lin edited this page Dec 15, 2023 · 3 revisions

Here describes the procedure to reconstruct FreeSurfer brain models using T1-weighted MRI data (MPRAGE sequence)

Set environment

Make sure that you have access to FreeSurfer software. SUBJECTS_DIR environment variable must be set.

$ setenv SUBJECTS_DIR /space_lin2/fhlin/eegmri_wm/subjects

Get the dicom data

FreeSurfer has a convention to put files in different subfolders, whjich will be created automatically. But the first sub-folder must be created manually. Here is my routine to reconstruct the brain for subject s004:

$ cd $SUBJECTS_DIR
$ mkdir s004; cd s004; mkdir mri; cd mri; mkdir orig; cd orig; mkdir dicom; cd dicom;
$ ln -s /space_lin2/kwoudsma/eegfmri_wm/s004/T1/* .

Basically, MPRAGE dicom files at /space_lin2/kwoudsma/eegfmri_wm/s004/T1/ were linked to $SUBJECTS_DIR/mri/orig/dicom.

A quick file listing gives:

...
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0154.2023.02.24.17.53.54.227018.233313898.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0155.2023.02.24.17.53.54.227018.233313916.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0156.2023.02.24.17.53.54.227018.233313934.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0157.2023.02.24.17.53.54.227018.233313952.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0158.2023.02.24.17.53.54.227018.233313970.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0159.2023.02.24.17.53.54.227018.233313988.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0160.2023.02.24.17.53.54.227018.233314006.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0161.2023.02.24.17.53.54.227018.233314024.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0162.2023.02.24.17.53.54.227018.233314042.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0163.2023.02.24.17.53.54.227018.233314060.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0164.2023.02.24.17.53.54.227018.233314078.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0165.2023.02.24.17.53.54.227018.233314096.IMA
EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0166.2023.02.24.17.53.54.227018.233314114.IMA
...

We learned that these are for run number 37.

Unpack dicom

Then, we "unpack" (convert) the dicom data into the FreeSurfer format. A unpack.rule file is created at $SUBJECTS_DIR/s004/mri/orig/unpack folder. The editor vi is used to create this file.

$ cd $SUBJECTS_DIR/s004/mri/orig/
$ mkdir unpack; cd unpack
$ vi unpack.rule

The content of unpack.rule file, in this case, is:

$ cat unpack.rule
37 3danat COR blah

This means that run 37 will be unpacked to 3danat folder in the COR format. The folder will be created during the unpacking process.

Once the unpack.rule is made, you can start unpacking.

$ cd $SUBJECTS_DIR/s004/mri/orig/unpack
$ unpacksdcmdir  -src ../dicom -targ . -cfg ./unpack.rule

If no error occurs, it will show the following.

...
Done scanning Wed Jul  5 15:37:53 EDT 2023
------------------------------------------
 37 t1_mprage_sag_p2iso1.0  ok  256 256 176   1 EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0001.2023.02.24.17.53.54.227018.233311136.IMA
unpacking config ------------------
{37 3danat COR blah}
-----------------------------------
---------------------------------------------------------
Run 37 -----------------------------------------
Wed Jul  5 15:37:53 EDT 2023
37 3danat COR blah
mri_convert ../dicom/EEGFMRI_WM_S004.MR.RESEARCH_FHLIN.0037.0001.2023.02.24.17.53.54.227018.233311136.IMA ./3danat/037 --sdcmlist ./3danat/037/flf -ot cor --nspmzeropad 3 --in_type siemens_dicom
 
 
StartTime: Wed Jul  5 15:37:41 EDT 2023
EndTime:   Wed Jul  5 15:38:11 EDT 2023
unpacksdcmdir Done


Convert COR to MGZ

Before running the FreeSurfer reconstruction, we need to convert again the COR files into MGZ format.

$ cd $SUBJECTS_DIR/s004/mri/orig
$ mri_convert unpack/3danat/037/COR-.info ./001.mgz

COR files in the folder $SUBJECTS_DIR/s004/mri/orig/unpack/3danat/037 are converted to 001.mgz` at $SUBJECTS_DIR/s004/mri/orig

Reconstruct the brain!

Finally, recon-all is all you need.

$ cd $SUBJECTS_DIR/s004
$ recon-all -subjid s004 -all

It will take about 10 hours to complete the whole processing.

Clone this wiki locally