Page MenuHomePhabricator

Incorrect RegionGrowing behavior on 4D data
Closed, ResolvedPublic


The issue is the following

  • Load a 4D dataset where the data is different for every time step
  • Perform region growing segmentation on time step 0
  • Notice that the segmentation is performed only in 3D and the result obtained for time step 0 is copied to the other time steps

The expected behavior would be to perform the region growing segmentation directly in 4D. In other words, the result should be different for every time steps, assuming different volume data.

Event Timeline

I managed to make the region growing work in 4D with minor code modification.

I would like to discuss the possibility to add the fourth dimension to the MITK_ACCESSBYITK_DIMENSIONS CMake variable. This is required in order to process 4D data with 4D algorithms.

Another way around it to use the AccessFixedDimensionByItk* macros and pass the current image dimension but I find this solution restrictive and somewhat ugly in terms of code.

What do you guys think?

I pushed my branch on Github [1] if you want to see/test the 4D region growing. All the tests pass so far.


Since there is already a fix provided by felix we should try to include it in the upcoming release.

Pushed bug-13999-IncorrectRegionGrowingBehaviorOn4DData
with a partial merge.
Some the PluginView has been deleted in the Segmentation Redesign,
when trying to adapt the changes to the new file, we run into the problem that implicitly the segmentation interpolation is used whcih does not work with 4D data.

Current release is finished. Reseting target milestone...

This bug is no longer present.