Slicer:Developer Meetings:20050201

From Slicer Wiki
Jump to: navigation, search
Home < Slicer:Developer Meetings:20050201

Slicer Developers Meeting

February 1, 2005, 2-3pm, at 1249 Boylston St., second floor demo room.

Topic: Randy Ellis gave a live demo of a commercial surgical navigation system developed as part of his research.

Present: Jonathan Sacks, Karl Krissian, Steve Pieper, Nicole Aucoin, Haiying Liu, Randy Ellis, Mike Halle, Katharina Quintus, Wendy Plesniak, Jeanette Meng, Sonja Pujol, Katie Hayes, Arne Hans, Faith Lin, +6 more

  • Randy is the newest faculty member at the SPL, specialises in orthopedic cases and is giving us a demo of a commercial package based on his research. It took $2M to produce the software after the spin off company was born, this talk focuses on software developement methodology.
  • focus is on fast development, example: 2 years from grant to having a case in surgery
  • they had a full time developer on the mission critical portions of the code, and students on the non mission critical portions
  • fail safe: hardware will fail (ie hard disk), test patient had a heart attack on the table, make sure that the surgeon can back out and recover from any failure
  • spin off company software development:
    • they re-engineered the software from the ground up, used VTK and the Qt widget set as they're safe, solid, multiplatform, Qt events generated by other software can be used to "poke" the system
    • used the extreme programming paradigm
    • UML designed around tests
    • link foot pedal to left mouse click, can do mouse jump in Qt to script the tests
    • managing bug fixes, regression tests
  • short demo of the software, hooked up to a tracker, bone with a "tumor" in it
    • register button - touch points on the bone
    • flouro button - 3D view, import real time scans
    • special views for ortho, pelvic, mesh
    • buttons, status indicators, undo/redo, calibration, help, interact with windows system
    • controller/model/world paradigm, each kept separate in the design as much as possible
      • problem is the interactions between them
  • someone developed a surgeon assist program that assisted in knee surgery using kinematics, no images Stryker
  • looked at workflow in the OR, they use very specialised tools and sets of tools that do one thing very well
    • storyboarded the process of ventrilcleostomy in CT scanner, produced a power point slide show that gave the state of the UI, one per slide. Start by identifying the target process, engineers watch how one is done now, record times, get problem points that they can clear up
      • initialisation
      • access data
      • load it
      • pin patient down in a head cage, rigid registration via fiducials
      • see internal structure that's between surgeon and ventricles
      • track problems
    • narrow down interactions to those that are necessary (task specific components only are used)
    • go from phantom to patient with IRB approval, surgeons is not to rely soley on the software (could test on animals, cadavers in the process as well)
  • Randy is the technical director for Amigo
    • he's canvassing clinicians for immediate needs
    • targetting TIPS (transjugular intrahepatic portosystemic shunt) as it used flouro, use ultra sound, patients have a preoperative CT angiogram
    • Amigo is the operating room of the future, will have 4 image sources in the OR, can move patients (slowly) along a track to a 3T magnet in one room, and then back. Can also move them into a PET CT in the other connecting room and back to OR.
    • going forward, break apart the 2 uses of the MRT
      • targetting - get where you want to be
      • monitoring - what changes have occurred
  • things to consider for bringing Slicer into Amigo
    • usuable in OR with multiple tracking technologies (ie endovascular ultrasound)
    • control software development
    • separate UI interaction from the controller
    • work flow - apply Randy's previous experience to developing some
  • avoid innovation and discovery in the OR, so have a linear work flow
  • he's got an interface to all Northern Digital trackers developed for the communcation package that he can distrubute, he prefers it over JHU CIS for software development history
  • Steve P
    • we'd like to keep the ability in the Amigo so we can have two software packages connected to trackers (suggested to GE Nav as well), so avoid serial ports
    • how do radiology techs follow the moving surgery? 1 or 2 more people are in the room, low numbers to keep chances of infection down. How control access near the MRT?
    • they will have multiple tracking technologies incorporated into Amigo
    • collaborative work is a goal, dynamic mrml updates



Back to meetings