XNAT Workshop 2012 - Pipeline Session

Goal:

At the end of this session, attendees should be able to:

• Add pipeline to a XNAT site
• Setup an available pipeline for a project
• Guide XNAT users on launching a pipeline
• Create a new pipeline

Video

Materials:

• Participants will need the workshop VM
• Dicom to Nifti Pipeline
• dcm2nii Resource descriptor
• XnatPipelineWorkshop2012 Powerpoint
• We will be exploring the folder /opt/xnat/pipeline_1_6 on the VM
• We will be looking at the DicomToNifti pipeline located at /opt/xnat/pipeline_1_6/catalog/mricron/DicomToNifti.xml. During the talk, we would add the pipeline to our site and launch the pipeline.
• Participants would be creating a pipeline which would dump the dicom headers into a text file.
• XNAT 2010 Pipeline talk (optional)

Prerequisites:

• Working knowledge of XML and XPATH
• If you have not deployed a XNAT instance, you should have attended talks on Day 1 (Project Intake, Project Configuration) and Day 2 (Data upload and entry)
• Working knowledge of linux commands.
• In a bash shell to redirect stdout of a command to a file, we use redirection character >
• In a XML document, > is a reserved character. To escape, we use &gt; for > and &lt; for <
• In a XML document, & is a reserved character. To escape, we use &amp;

Tasks during the talk:

Step 1: Adding a pipeline to a site

• • • We would add the pipeline /opt/xnat/pipeline_1_6/catalog/mricron/DicomToNifti.xml to the site

• • • Login to XNAT as Admin.

• • • Navigate to Administer -> Pipelines
    • Click on Add Pipeline to Repository
    • Input
      
      • Enter Path to Pipeline descriptor xml:   /opt/xnat/pipeline_1_6/catalog/mricron/DicomToNifti.xml
      • Enter Name of the custom webpage to launch this pipeline: LEAVE THIS BLANK
      • Click on Add
      • You will get a message "... was added successfully
      • If you did everything correct, you will see DicomToNifti listed in the pipeline repository.

Step 2: Setting up the pipeline for the project BOGUS_OUA

              Navigate to BOGUS_OUA project as project owner

1. • Click on Pipelines tab.
     
   • Click on Add More Pipelines
   • Click on Add (for DicomToNifti pipeline)
   • Select Is this pipeline to be launched automatically while archiving?
     
   • Set the value of create_nii to N
   • Click on Submit
     
     
     

Step 3: Setup the datatype

1. • As Admin, click on Administer -> Data types
   • Select xnat:mrSessionData
   • Under Available Report Actions, enter
     
     • Name:  PipelineScreen_launch_pipeline,
     • Display Name: Build,
     • Popup: always,
     • Secure Access: edit
     • (leave other columns empty)
       
       

Step 4: Launching the pipeline (via Action -> Build)

1. • Navigate to project BOGUS_OUA

   • Select an MRSession
   • Click on Action -> Build
   • Select DicomToNifti
   • Click on Submit

Step 5: Launching the pipeline automatically

1. • Upload a MR Session

   • Archive the session

Step 6: Provenance information

1. • Navigate to /opt/xnat/data/build/BOGUS_OUA/DATESTAMPED_FOLDER/SESSION_ID_HERE

Step 7:Create a new pipeline - DicomDumpToText = HELLO PIPELINE WORLD!

Problem Statement:

Write a pipeline which will create a text file containing the dicom header dump. 

Assumptions:

• The input scan files are DICOMs
• The file names are *.dcm
• The generated text file will be a resource at the SCAN level

Executable: You could use any. You could use dcmdump or dcm_dump_file.  If you want to use  dcm_dump_file, it  is at /opt/xnat/workshop2012/d3s1-pipeline/pipeline/DicomDumpToText/resource

Process:

Step a) [Skip this if you are using dcmdump] Copy dcm_dump_file to /bin: sudo cp /opt/xnat/workshop2012/d3s1-pipeline/pipeline/DicomDumpToText/resource/dcm_dump_file /bin

Step b) [Skip this if you are using dcmdump] sudo chmod +x /bin/dcm_dump_file

Step c) Create a resource descriptor for dcm_dump_file/dcmdump

            Hints:

• • dcm_dump_file *.dcm > scanID.txt OR dcmdump *.dcm > scanID.txt

                     <pip:argument id="redirect_stdout" prefix="">

                        <pip:name>&gt;</pip:name>

                       <pip:description>Redirect Std Out</pip:description>

                   </pip:argument>

• • Use /opt/xnat/pipeline_1_6/catalog/mricron/resource/dcm2nii.xml as a template
  • REST URI: PUT to

         /data/archive/projects/PROJECT_ID/subjects/SUBJECT_ID/experiments/XNAT_ID/scans/SCAN_ID/resources/DICOMDUMP/files/SCANID.txt?format=TXT&content=DICOM_DUMP

Step d) Create a pipeline XML. Use  /opt/xnat/pipeline_1_6/catalog/mricron/DicomToNifti.xml as a template                    

Step e) cp -r /opt/xnat/workshop2012/d3s1-pipeline/pipeline/DicomDumpToText /opt/xnat/pipeline_1_6/templates/misc/catalog

Step f) cd /opt/xnat/pipeline_1_6; setup.sh xnat@xnat.org mail.nrg.wustl.edu http://localhost:8080/xnat

Step g) Add pipeline to site

Step h) Add pipeline to project

Step i) Launch the pipeline

Solution: /opt/xnat/workshop2012/d3s1-pipeline/pipeline/DicomDumpToText

More:

1. REST URI to look up the archive specification for a project:
   http://localhost:8080/xnat/REST/projects/BOGUS_OUA/archive_spec
2. dcm2nii -
   http://www.mccauslandcenter.sc.edu/mricro/mricron/dcm2nii.html
   
3. Location of builddir - /opt/xnat/data/build
4. Script to overload to interact with a Grid - /opt/xnat/pipeline_1_6/bin/schedule