Slicer:Developer Meetings:20070612

From Slicer Wiki
Jump to: navigation, search
Home < Slicer:Developer Meetings:20070612

Time: 2pm, 1249 Boylston St., 2nd floor alternate meeting room, Boston, MA.

Agenda

  • PNL wishlist for Slicer 3

Attendees

  • Steve Pieper
  • Katharina Quintus
  • Nicole Aucoin
  • Katie Hayes
  • Sylvain Bouix

Minutes

Sylvain Boiux, Katarina Quintus, Steve Pieper, Nicole Aucoin, Katie Hayes

  • PNL group needs to use Slicer3 to give some feedback, but not for something on the critical path
  • Tri Ngo's stuff? DTI Stochastic Tractography
  • EM segment
    • registration is still missing
    • they use the EM Atlas Brain Classifier in Slicer 2.x
  • Editor needs to be tested to check how it stacks up against the Slicer 2.x version, what's missing, what's better

To transition to Slicer3, they have wishlists of functionality.

Structural scans:

Urgent needs:

  • EM Segmentation
    • rigid transform T2 to SPG
    • non linear registration of atlas to subject
    • classification
    • Brad Davis working on this
  • Manual drawing
    • on brain mask from auto segmentation to make corrections
    • Steve P will continue work on Editor
  • Realignment
    • landmark based
    • can be done with 2d fiducials, on the slices
    • can be done as one call to a CLI, pass fiducials in, transform out
    • ping Jim Miller about getting transforms out of CLIs Jim's Slicer3 list
    • fiducials needed: one list along the mid sag, 2 points for fixing yaw
    • Nicole will code up a CLI
  • Slices mode (similar to the slicer2 Transform Volume Module -- allows you to apply a transform to a set of volumes; can specify target orientation, spacing, dimensions)
    • Sylvain uses this all the time
    • does fit to window come close?
    • Steve P.

Slightly less urgent needs:

  • Resample as a CLI
    • may be difficult
  • Volume statistics
    • interactive?
    • being able to see statistics when peek under a label map will be enough to start
    • Katharina will work on that
    • itkLabelStatisticsImageFilter should be able to provide most of the intensity based information
    • Dirk Padfield will be working on an itkGeometryLabelStatisticsImageFilter to calculate more geometric features
  • ROI drawing
    • In GE-land, this means a rectangle, used to define subvolumes, bounding boxes, pass to CLI
    • In PNL-land, this means a label map, drawn in the editor
    • Editor needs to suport paint, draw
    • CLI for dilate/erode, remove islands, threshold
    • Ablility to save and reapply drawn areas on the next slice
  • Registration
    • need it to give up a transform
    • what types of transforms to support? Linear? Rigid? Non-linear?
    • interactive in that the initialisation (translation) can be manual before set it off auto, with slices
  • Transform volume
    • pass in a transform
    • must be able to run in batch mode, so can transform 90 volumes easily
  • ModelMaker has to work
  • Testing
    • Brad is writing regression tests for his own code in EMSeg
    • be aware that baseline images from Slicer 2.x won't match those from Slicer3, as Slicer3 doesn't do the flip in Y (but any NRRD files should be identical)

The wishlist for diffusion is less urgent, as the infrastructure is still being worked on (can load and save volumes, but can't really look at them all that well, nor generate tracts).

  • measurement frame bug fix is necessary in the NRRD reader/writer
    • add a known bugs item in the 2.6 documentation, fix before the 2.7 release

Tutorials:

  • Sonia is working on Slicer3 ones
  • has a preliminary developer on for making command line modules
  • hopefully done by end of summer, but summer students in PNL can help test, feedback, and maybe even write segments

Feedback session

  • second week in July will have a hands on tutorial/feedback session covering the editor
  • PNL will nominate attendees to give us feedback: Toshiro, Charlie Yu, plus a newbie
  • Todo for that date:
    • realign
    • extract and apply transforms