IGT:2005 Nov 16 IGT Consortium Kickoff

Meeting Agenda

1. Introduction of Projects (getting to know each other) 10:15-2:00

Each party presents Example Applications, Design, Requirements, Codying Style, Dashboard, Choice of linked tool, License, Support, etc

Each party is expected to upload PPT files (and/or their handouts in PDF) after the workshop.

• IGSTK (Kevin Cleary)(10:15-11:00)
  • File:IGT-Meeting-Nov16-IGSTK-Architecture-Overview.ppt
  • File:IGT-Meeting-Nov16-IGSTK-StateMachine.ppt
• File:CADDLab-IGS.ppt (Julien Jomier & Stephen Aylward) (11:00-11:45)
• CISST-MRC (Peter Kazanzides)(11:45-12:30)
  • File:CISST-Software.ppt

• LUNCH (from Brown Sugar Cafe Thai Food) Please look at the lunch menu for the Fenway location and be ready to place your order by 10:30)

• Slicer-Navigator: incl. History of IGT at BWH, Current State of MRT research, Dr. Jolesz's priority cases in AMIGO

(Randy Ellis)(1:00-1:30)

• GE Relationship (Tina K) (1:45-2:15)

- equipment shipments/installation - perspectives on the software infrastructure

2. Discussion (Clarifying the difference/overlap) 2:00-3:00

Chair: Randy Ellis

Participants go through various aspects of IGT software and find difference/similarity in their approach.

The chair of the session has summarized the discussion (below).

• License
• Dashboard
• Linked library (numeric library, interface tool kit)
• Thread-safety
• Communication with external programs

- Integration of new trackers/instruments/

• Architecture

- State machines - Integration of various toolkits

• IRB
• Requirements for FDA Platform (i.e. nav system)

3. Collaboration (Finding the complimentary/collaborating relathionship)3:00-3:30

Open discussion

4. Action Plan (what do we do next?)3:30-4:00

• Workshop in March, 2006 (Ron to work with Kevin Cleary on 2-page proposal with the group)

Logistics

Place/Time

10:00am - 4:00pm

at SPL,1249 Boylston Street, Boston MA

Direction -> Splweb.bwh.harvard.edu:8000/pages/directions/1249Boylston/Boylston-facility.html

Lunch

Look at menu (Thai food link above on this page). Place order by 10:30.

T-CON Instruction

Dial 1-218-936-1100 and enter the Conference ID 43123 followed by # key. If conference is not in session, system will put you on hold until the moderator arrives. During the Conference

• 3 - increase volume on the line
• 6 - mute individual line
• 7 - un-mute individual line
• 9 - decrease volume on the line

Attendees (please add your name, available time, method of attendance)

• Nobuhiko Hata (BWH), not available to attend, in flight from Tokyo to LAX
• Tina Kapur (NAMIC, BWH IGT), available at 1249
• Steve Pieper, at 1249
• Julien Jomier (CADDLab), at 1249
• Stephen Aylward (CADDLab), TCon only
• Kevin Cleary (Georgetown), tcon, not available 11 am to 1 pm
• Peter Kazanzides (JHU, ERC-CISST), attend
• Patrick Cheng (Georgetown), attend
• Ron Kikinis
• Luis Ibanez (Kitware), attend at 1249
• Will Schroeder (Kitware), attend at 1249
• Rick Avila (Kitware), attend at 1249
• Bill Lorensen
• Ferenc Jolesz
• Mike Halle
• Jim Miller
• Xiaodong Tao
• Sandy Wells

Summary Meeting Notes

• License details:
  • General agreement that BSD-style license are preferred
  • GPL-LGPL details were rehashed -- a continuing discussion...
• Dashboard: Most eagerly await DASH2, all agreed that dashboards

are an important part of software QA (during both development and maintenance phases of software development)

• Libraries: general agreement on sharing of VTK/ITK, we noted
  • JHU vector/matrix classes have efficiency advantages
  • JHU robotics interfaces are best available
  • Overlap of JHU and Kitware NETLIB wrappers/translations
  • Significant divergence of tracker libraries: JHU has CISST classes, IGTSK has other classes, BWH is inclined to use OpenTracker libraries (an inevitable consequence of independent needs of a critical software resource)
• Thread safety raised two matters:
  • Legacy code is being ported as resources permit (Kitware is tasked by Terry Yu of NLM to decontaminate NETLIB)
  • All are aware of thread safety in development of new code, with proviso that this needs sophisticated developers
• Communication with external programs raised lively debated:
  • The Slicer Navigation system must communicate with Slicer 3.x processes that run "hard, non-realtime" computations that support interventional applications
  • Numerous publish-subscribe mechanisms were mentioned, the chair (REE) saw no consensus on solutions within this evolving space
  • Much discussion of CORBA, which is clearly necessary for interaction with GE software (especially navigation)
• Architecture was a subliminal topic of the day:
  • No explicit discussion
  • Consensus was for layers: build on ITK/VTK, have toolkits such as IGSTK provide software base, and then applications (i.e., interventional procedures) in distinct "silos" that solve specific clinical problems
  • State machines were identified as a particularly apt abstraction of task-based descriptions of interventional procedures
• IRB (Institutional Review Board)
  • Identified as the target audience of the quality-assurance process of sn interventional software/system
  • We noted the complex relationship of the IRB, patients consenting to an investigational treatment, and the discretion of physicians who may treat patients using their clinical judgment based on the totality of information available at the time of treatment
• FDA (Fooda nd Drug Agency): consensus was that we are following good manufacturing processes in our development and testing phases; discussion on processes for elucidating software design processes and bug issues (reportin, correcting, disseminating fixes, interim warnings, etc.) raised topics for future discussion