CIBC:Documentation:Seg3D:FeatureRequests:March8-07-List

A list of features for Seg3D that from March 8, 2007. These items started with Josh and McKay and were then prioritized by the management and exec teams.

• Important
  • Modifications to the layer dialog. (In progress) (Benefits: NCMIR)
  • Modify GUI to use wxWidgets. As a first step, we could simply replace the file browser and menus with wxWidget widgets. (Benefits: all)
  • Documentation: need tutorials. (Benefits: all)
  • Documentation: need a user guide. (Benefits: all)
  • Add ITK Active contour segmentation. Requested by Ruth and Iman at NCMIR. The infrastructure is in place (level-set segmentation), but we need to sort out how to load training data and what the GUI might look like. Iman would like to do a project using this algorithm. (Benefits: NCMIR) (COmment: this is a lot of work and we should try and find a student who needs it.)
  • Save output of isosurfacing to VRML/Open Inventor format, OR otherwise figure out how to import Seg3D output into Amira. This would be very useful to the NCMIR group. (Benefits: NCMIR) (Comment: this is important; need to do this to make it useful. We should save in VTK format and let them worry about how to convert to other formats.)
  • Input to Seg3D: we need to do some work on this and identify needs. Whatever we do should be general for all our apps. Check and document what we have now via ITK readers.
  • Add additional ITK filters: Bryan Smith, postdoc at NCMIR, requested a threshold filter followed by connected component labeling. This would be easy. In general, we need more ITK functionality in Seg3D. A couple of weeks where we just add ITK filters and work on their Seg3D dialogs would be nice. (Benefits: all)(Comment: straighforward and moderately useful. Might be a good project for Sila)
  • Develop mapping of OWL ontologies (only a few concepts) to binary logic for label masking. Design a simple interface for saving/loading ontology info. (Benefits: NCMIR) (Comment: we need a clear plan before we do anything; this could consume a lot of resource so we have to be careful. Review plan with NCMIR. Would be useful to multiple groups)
  • Color Lookup Table editor - specifically the ability to modify colors mapped to out of range values (0, or 255) in order to quickly identify saturated pixels in otherwise grayscale images. (Benefits: all) (Comment: make sure this is useful from user perspective. Is this the same sort of thing as XV? If so, it would be very useful. Is there a more general solution for color map editing that we could develop, using WxWidgets.)
  • Documentation / Code refactoring to make adding new ITK filters easy for contributing developers.
  • Copy a previous (or next) slice label (2D) into the next slice (or previous) using k, cnt-k. This was requested by NCMIR for manual segmentation. (Comment: Ross thinks we already had this! It should be there!!)
  • Save current 3D view as PNG. Requested by NCMIR. (Comments: this is easy so probably we should do this, perhaps using Teem support for PNG)

• Less Important
  • Simplify the filter dialogs, i.e. the basic interface for threshold-based ls segmentation might simply use 2 sliders, one for curvature and one for std. deviations from the mean. An advanced menu could also be available which exposes all the ITK parameters. The NCMIR users might be more successful with the ITK segmentation filters if we could make their use more intuitive. (Benefits: naive users)
  • Flood fill outlines in 2D for faster manual segmentation (a user may specify just the outline of a region to fill). Requested by NCMIR. (Comment: Not clear what this means. It seems like we should have this capability already present. If they mean we should be working with polygons, then "no")
  • Develop strategies for working with very large datasets. A file reader that downsamples data as it reads would help with this: full-resolution subregions overlaid on a low-res, global context would be a nice way to process datasets in manageable chunks within seg3d. In general, multiscale schemes for processing data should be considered. (Comment: we have been moving in this direction and should think about this. We do need to work on this and see which parts are in place. Would be a great project for a student to take on. )

• For Longer-term consideration
  • ITK affine registration. Requested by NCMIR postdoc. (Comment: big problems and ones we need more support to achieve)
  • ITK deformable registration (2D). Requested by Gina Sosinsky at NCMIR for stacks of cryosection data. Potentially benefits Elliott McVeigh. (Comment: big problems and ones we need more support to achieve)
  • Simple Camera Path editor for generating movies. Requested by NCMIR. (Comment: this is hard and we should not go near this black hole. There are proprietary solutions available for this. We might find a simpler solution).

Go back to Documentation:Seg3D