IGT: Workflow: TCon

Back to IGT Workflow Breakout Page

Administration

• Tuesday, Nov 28th at 3pm EST
• Attendees
  • Stephen Aylward, Steve Pieper, Luis Ibanez, Kevin Cleary, Tina Kapur, Brad Davis, Sebastien Barre, Nobuhiko Hata, Eigil Samset, Simon DiMaio, Jim Miller, Bill Lorensen, Will Schroeder, Terry Yoo,
• Telecon info:
  • Dial-in Number: 800 569-2474
  • Participant Pass Code: 574104

Proposition

1. A task-level workflow framework, that is GUI independent, should be added to ITK.
   1. Lorensen asks: Should this be in ITK? Can we do something that is toolkit independent?
   2. Lorensen asks: Is there an existing workflow engine that we can use?
2. Companion methods for recording, templating, simulating, visualizing, and analyzing workflows should be developed using that framework.
3. A workflow file format should be defined.
4. Slicer, IGSTK, SIGN, and VolView can then build upon that common framework and benefit from the companion methods.

Background

• The focus is task-level workflow, not data-level workflow (e.g., not ITK's pipelines).
• Imaging projects with workflow
  • IGSTK (Cleary/Ibanez - workflow for IGT)
  • SIGN (Samset/Hata - workflow for IGT)
  • Slicer3 (Davis/Barre/Ibanez - workflow for Slicer modules)
  • Kitware/AFRL (Schroeder/Aylward/Ibanez - workflow for segmentation)
• Imaging events discussing workflow
  • Follow-on to 2006 IGT Workshop: breakout session on workflow
• Standards in workflow
  • Visual programming using the Object Constraint Language (OCL)
    • Umbrello from the KDE project
  • File formats
    • XMI (from the Object Management Group that defined the UML standard)
  • Specification
    • UML2's Activity Diagrams
      • [Dumas 2001]

Review IGT Breakout recommendations

1. Create a library of reference workflows
   • Include clinicians in the process
   • Storyboards and timings
   • Data from Stealthlink/VectorVision
2. Realize that workflows should be adaptive
   • Specialize for site, physician, patient, problem
   • Flexible
3. Workflows should be a basis for validation
   • Tune an IGT system / compare systems
   • Tune a clinical procedure
   • Workflows should carry accurate/time estimates as they progress
4. Workflow execution should be GUI independent
   • Okay to place some burden on developers
   • Different execution models exist: Lemke / service oriented architecture
5. Data workflows (processing a large number of images) should also be addressed

Goals of tcon

1. Build consensus for target implementation
2. Commit to contributing to and using target implementation
3. Identify tasks to reach the target
4. Accept volunteers for tasks

Status reports

• IGSTK
  • Workflow Validation
  • Application Builder
    • UML2, OCL, and XMI
• SIGN
• Slicer
• AFRL

Tasks

1. Separate lib or ITK lib
2. Recording tool
   • Template and timers for recording in the OR
     • Speech recognition
     • Audio recordings
   • Create library
   • Distill to a few templates
3. Execution model
   • Use (extend) existing standard
4. Testing tool
   • Test ex vivo
     • Kevin Gary
   • Test in vivo
     • Workflow as the language/basis for testing
     • Issues events
5. File format
   • Support templates
   • Use existing standard

tcon notes

1. Library should be independent of ITK/VTK/KWWidgets
   • Perhaps distribute it with those toolkits in their Utilities directories.
2. A small number of templates can represent most interventional radiology workflows [Stephen is looking for the reference]
3. Regarding IGSTK
   • State machines are used
   • People have concerns about appearance of complexity
     • Luis says such complexity is in the surgical procedure - not created by state machines
     • A strength of state machines is that they expose this complexity
   • Another strength of state machines is determinism / order-of-operation
   • State machine engine of IGSTK has been ported to a set of classes in KWWidgets and used in a Slicer 3 module
     • Seb Barre did the state machine port under guidance from Luis.
     • Seb added features and a wizard workflow engine.
     • Seb made it so that it could be easily extracted to an isolated library. Update (Seb): I do actually take that back. We wanted KWWidgets to be independent of VTK as well back in the days, but we had to backtrack given the amount of code and features that are brought in even by simply subclassing from vtkObject. This is open to discussion, but not easy.
     • Should be ported back into IGSTK. Update (Seb): I'm not sure what you mean by that?
4. Regarding VTK
   • InfoViz project is developing graph/state-machine viewers
5. Regarding SIGN
   • The project uses the Slicer MRML data foundation and state machines
   • Work is ongoing
     • Current emphasis is visual programming workflows
     • Provides support for parallel workflows
       • Parallelization = activate/deactivate workflow elements in parallel
       • Parallelization may also be available in Seb's implementation
   • Documentation and source on webpage is current (within 1 month)
6. Regarding Amigo
   • Yall project is meant to capture workflow
     • Lead by Michael at BWH
   • Noby distinguishes workflow editing, logging, and guidance
     • Right now they are focusing on editing
7. Regarding other players
   • University of Delft
     • Working with MeVis
     • Interviews and recordings to capture workflows
   • HU Lemke of CARS fame has been a strong proponent of workflow-centric research in IGT
     • | Google search on Workflow and Lemke
     • See also http://www.iccas.de/index.php/59/0/
     • From Oliver Burgert from Dr. Lemke's group: At a very high level, one may say that Leipzig has focused on tools for

workflow analysis so far and what we should look at is tools for workflow implementation of image-guided systems for interventional procedures

1. #* #** Kevin Cleary has a strong connection with them
   • • A paper on workflow recordings in interventional radiology with Drs. Lemke and Cleary is here.
2. Regarding workflow vocab
   • From Simon DiMaio: "The discussion reinforced for me the apparent gap/disconnect between the concept of clinical task-level workflows, algorithm task-level workflows and data-workflows. Folks tended to become bogged down in the space where task-level workflows meet data-level workflows. Marrying the two in an elegant and reusable may be the grand challenge. This may also be where creating an independent class library, or such, could run into difficulty."

tcon recommendations

1. Library should be independent of ITK/VTK/KWWidgets
2. A vocabulary for workflow needs to be defined
   1. What is meant by workflow
3. Establish a repository for workflow "recordings"

References

• [Allen 1983] Allen JF, “Maintaining knowledge about temporal intervals.” Communications of the ACM, 26(11) 1983 pp. 832-843
• [Brief 2005] Brief J, Edinger D, Hassfeld S, Eggers G, “Accuracy of image-guided implantology” Clinical Oral Implantation Research, 16 2005 pp. 495-501
• [Dickhaus 2004] Dickhaus CF, Burghart C, Tempany C, D’Amico A, Haker S, Kikinis R, Woern H, “Workflow modeling and analysis of computer guided prostate brachytherapy under MR imaging control” Studies in Health Technology and Informatics 2004; 98:72-74
• [Dumas 2001] Dumas M, ter Hofstede AHM, “UML Activity Diagrams as a workflow specification language” Proceedings of the UML 2001 conference
• [Jalote-Parmar 2006] Jalote-Parmar A, Pattynama PMT, Goossens RHM, Freudenthal A, Samset E, De Ridder H. "Exploring a user centric methodology to investigate and integrate information gathered during medical intervention." In Proceedings of the 16th World Congress on Ergonomics, Maastricht, The Netherlands: International Ergonomics Association. 2006.
• [Kolski 2002] Kolski C, Vanderdonckt J (eds), “Computer-aided design of user interfaces III.” Proceedings of CADUI 2002, Kluwer Academics, Dordrecht, 2002
• [Paggetti 2001] Paggetti C, Martelli S, Nofrini L, Vendruscolo P, “Interface design and evaluation for CAS systems.” MICCAI, 2001 pp. 1099-1106
• [Paterno 2000] Paterno F, “Model-based design and evaluation of interactive applications.” Springer-Verlag, Berlin 2000
• [Raabe 2003] Raabe A, Krshnan R, Seifert V, “Actual aspects of image-guided surgery.” Surgical Technology International, 11, 2003 pp. 314-319
• [Trevisan 2002a] Trevisan DG, Raftopoulos C, Macq B, Vanderdonckt J, “Modeling Augmented Reality System, Image Guided Surgery Case Study.” Proceedings of the 1st Glasgow Context Group Colloquium GCG, 2002
• [Trevisan 2002b] Trevisan DG, Vanderdonckt J, Macq B, “Analyzing interaction in augmented reality systems.” International Workshop on Immersive Telepresence, 2002
• [Trevisan 2003] Trevisan DG, Vanderdonckt J, Macq B, Raftopoulos C, “Modeling interaction for image-guided procedures.” SPIE Medical Imaging, 2003

Back to IGT Workflow Breakout Page