Deleted branch from rMITK MITK: release/T28750-2021-Week-42.

Pushed new branch to rMITK MITK: release/T28750-2021-Week-42.

I guess in GIMP the behavior is somewhat similar to your second suggestion. As you are not able to set additional points, it is a "restricted region". You can still modify existing points. This is what the tool looks like in GIMP if you move one of the existing points.

One comment to this: The tool was inspired by the behavior of a similar tool in GIMP. In GIMP it is not possible to add additional points after finishing the contour. So they don't have that problem.

The users have 2D+t data. The current workaround is to convert the 2D+t data to a 3D volume, where one slice is one timepoint and interpolate like this. This is not a perfect solution, as the 3D interpolation tries to interpolate to something like a "round" or "organ shaped" object, which does not make sense in this case.

I will close this now, discussion can be continued in T28741.

Thanks!

In T24358#228955, @kislinsk wrote:

This is a good candidate for some XML parsing errors that we see in the latest develop version of the MitkWorkbench, For example, activate any 3D segmentation tool and look at the console output:

vtkXMLParser.cxx:379 ERR| vtkXMLParser (0000027698218908): Error parsing XML in stream at line 1, column 0, byte index -1: no element found

It starts with line 162 in mitkTool.cpp (last code line in snippet):

auto displayActionEventBroadcast = dynamic_cast<DisplayActionEventBroadcast*>(us::GetModuleContext()->GetService<InteractionEventObserver>(*it));
if (displayActionEventBroadcast != nullptr)
{
  // remember the original configuration
  m_DisplayInteractorConfigs.insert(std::make_pair(*it, displayActionEventBroadcast->GetEventConfig()));
  // here the alternative configuration is loaded
  displayActionEventBroadcast->AddEventConfig(m_EventConfig.c_str()); // line 162, m_EventConfig is an empty string
}

I guess m_EventConfig is not supposed to be an empty string. Or, if it is supposed to be an empty string, AddEventConfig() should probably handle it gracefully, before the vtkXMLParser has the chance to complain.

This is a good candidate for some XML parsing errors that we see in the latest develop version of the MitkWorkbench, For example, activate any 3D segmentation tool and look at the console output:

Thank you for such a quick answer. I have now got MITK to build with clang.

I found one last hint, about a possible interference with ccache , last try ...

I tried several things again, similar to described above, same situation.

Hi and thank you for the comprehensive report. I noticed you use GCC. We support Apple LLVM (Clang) on macOS Mojave, Catalina, and Big Sur. See https://docs.mitk.org/nightly/SupportedPlatformsPage.html for support level details regarding tier 1 and tier 2. Our continuous and nightly build reports can be found here: https://cdash.mitk.org/index.php?project=MITK.

The labels in my example are consecutive integers. And even if they weren't the color mapping should skip unused integer values. Use equivalent of np.unique(segmentation) to determine used labels and set the map accordingly.