Mbirn: BIRNDUP
GENERAL INFORMATION
- Goals and Participants
- BIRN De-identification Pipeline tutorial
- BIRN De-identification Pipeline: Technical Report (Jul 2004)
- BIRN-DUP use instructions for installation at MGH
UPDATES
March 7, 2006 (S. Pieper)
Started a new page about BIRNDUP Distribution with instructions for the cvs repository.
The new repository includes the atlases and executables needed to run the process on linux.
March 2, 2006 (S. Pieper)
Work progressing on porting BIRNDUP user interface to Slicer 2.6 and integrating new scripts from MGH. Open items include:
- new cvs repository to replace lost nbirn.net code (BWH)
- need staticly compiled executables for mri_deface and related code (MGH)
- streamline UI, improve rendering (BWH)
- data provenance added to data structures (BWH)
- upload scripts for XNAT, HID (WashU, UCSD)
February 10, 2006 (K. Helmer)
S. Czanner defaced the PhaseII data. It went quite smoothly on seychelles. You can check the results in: /autofs/space/dijon_021/users/czanner/DEFACED_DICOMS/PhaseII
February 8, 2006 (K. Helmer)
S. Czanner now has a new version of birnd_up script, which is freesurfer installation independent:
/autofs/space/dijon_021/users/czanner/BIRND-UP.tar.gz
February 8, 2006 (K. Helmer)
D. Greve modified mri_deface so that it does not (should not) need a license. Can someone test it when it's live tomorrow?
February 7, 2006 (K. Helmer)
S. Czanner created a new script for anonymizing and defacing dicom files:
I used script dcanon.tcl from Slicer and mri_deface from Freesurfer.
You can get the whole package from: /autofs/space/dijon_021/users/czanner/BIRND-UP.tar.gz
I tested this script on seychelles and on my machine gabi. An example of using this script is in: /autofs/space/dijon_021/users/czanner/DEFACED_DICOMS/Harvard/scripts/seychelled_deface.csh
I used data from Stuart Wallace: /autofs/space/mojave_002/users/swallace/mind_reliability/dicom/structural/harvard
There are 10 subjects, each subject has 2 visits, each visit has T1 and T2 volumes. First T1 was defaced, then T1 and T2 were registered. Based on this registration, a mask from T1 was created. Then T2 was defaced using the mask from T1. The anonymized and defaced volumes are in: /autofs/space/dijon_021/users/czanner/DEFACED_DICOMS/Harvard
Command Line Version Usage
BIRND-UP USAGE
To run the script properly, you need to setup the following two variables:
Setup the environment variable BIRND_UP_DIR to the BIRND-UP directory. For example:
setenv BIRND_UP_DIR /my_home_directory_path/BIRND-UP
With these environments you can define aliases to run the scripts:
alias birnd_up $BIRND_UP_DIR/scripts/birnd_up
And then just run the scripts by typing:
birnd_up (or birnd_up -help)
How to use it
USAGE: birnd_up [options] -i <indir> -o <outdir> -subjid <subjID>
-i indir : input directory <dcmdir>
-o outdir : output directory
-study studydir : study directory
-subjid subject_ID : subject ID
-radius radius : radius in [mm], default 5
-radius distance_from_brain : Default is 7mm, but if a bit of brain is being stripped, it can be increased, or if too much is face is left, it can be decreased.
-result_check : create tif images from defaced volume
-version : print version string
-help : show the usage text
Examples:
set DATA1=/your_path/data1
set DATA2=/your_path/data2
set OUTPUT_DIR=/your_path/defaced_dicoms
./birnd_up -i $DATA1 $DATA2 -o $OUTPUT_DIR -subjid test_subj
Defaced dicom files will be in:
$DATA1 --> $OUTPUT_DIR/test_subj/scan_1 $DATA2 --> $OUTPUT_DIR/test_subj/scan_2
./birnd_up -i $DATA1 $DATA2 -o $OUTPUT_DIR -subjid test_subj -study1 visit1
Defaced dicom files will be in:
$DATA1 --> $OUTPUT_DIR/test_subj/visit1/scan_1 $DATA2 --> $OUTPUT_DIR/test_subj/visit1/scan_2
./birnd_up -i $DATA1 $DATA2 -o $OUTPUT_DIR -subjid test_subj -study1 visit1 -result_check
Defaced dicom files will be in:
$DATA1 --> $OUTPUT_DIR/test_subj/visit1/scan_1 $DATA2 --> $OUTPUT_DIR/test_subj/visit1/scan_2
TIF images from defaced data will be in:
$DATA1 --> $OUTPUT_DIR/test_subj/visit1/scan_1_tif $DATA2 --> $OUTPUT_DIR/test_subj/visit1/scan_2_tif
November 18, 2005 (S. Pieper) Modifications were made to the dcanon process to accomodate a newer version of the mri_deface program and related updated atlases. These updates were used to anonymize datasets made available at the SfN meeting.
October 3, 2005 (S. Pieper) The last phase of the BIRNDUP pipeline (the upload to BIRN database) has been a pending item dependent on the computational informatics groups. The BIRNDUP developers have decided to make this part modular, with BIRNDUP outputing defaced and deidentified DICOM images to a local directory for later upload with a specialized script. The current goal is to have a version of the program meeting this revised spec available for dissemination at SfN.
July 27, 2005 (S. Pieper) New BIRN Portal application distribution page will be an ideal place to make BIRNDUP available to the community (since it provides registration and download tracking). This is being researched.
May 17, 2005 (J. Jovicich) Essentially on hold until defacing validation is complete. It has been tested at other sites with T1-weighted data and it worked fine. Amanda tested on T2 data and it failed, but using a mask from a T1-scan acquired at the same session it worked. Distribution of the mri_deface binaries and related atlases will not be an issue. We are in the process of trying to distribute source code as well, and it looks promising. (J. Jovicich)
Feb 3, 2005 (S. Pieper)
- BIRNDUP is being installed and tested at Dartmouth (Brain Imaging Lab, Andy Saykin) to meet their IRB requirement for facial deidentification of SPGR scans for NA-MIC.
- Some installation troubleshooting on the full BIRNDUP interface
- The mri_deface and header deidentification scripts appear to work fine on their SPGR GE Dicom files
Jan 11, 2005 (S. Pieper)
Update regarding recent work on the upload script, which may affect the plans for BIRNDUP. Last year mBIRN and fBIRN Informatics created a joint SRB hierarchy, which will be used for both structure and function scans. Recently, the fBIRN Informatics group has created an upload script corresponding to this new hierarchy. It will upload both mBIRN and fBIRN data transparently. Burak's upload script which was used for Phase 1 of mBIRN is no longer operational. I understand that the new upload script will be placed in BIRN CVS very soon, certainly before the next fBIRN upload starts in February. BIRNDUP is not required for this upload, but will be need to be included in the upload pipeline in future. Also, you should also be aware that there are fBIRN sites which may require defacing of T2 scans, depending on IRB requirements. This is not an immediate requirement. Contact either Dave Keator or Dingying Wei
External Links
DICOMAnonymizer
A similar free java-based tool is here: http://eng.neologica.it/download/downloadDICOMAnonymizer.html Note that this program:
- does not remove facial features
- only handles one dicom file at a time
- cannot be run in batch mode
The professional version ($39) removes some limitations, but still doesn't address defacing.
Back to Segmentation Tools
Back to Mbirn: Analysis, Visualization and Interpretation Tools
Back to Mbirn:Main_Page