araex / mitk-gem Goto Github PK

I got into a state where it got stuck loading .vtu. The GUI is still somewhat responsive, but the progress bars shows 0% and you can't really do much. Restarting the app did not resolve the issue. However, it only appears to be happening when i try to open only a .vtu. Opening the same .vtu together with a regular image (as in 2 files at once with multiple selection), it works for some reason?

Currently it contains the descriptions from the tetgen documentation. Some more explanation would be helpful.

-a has a default value computed using mesh volume information. This should be explained as usually this is a constant value.

CPack messes up some of the dependencies of the UGrid Visualization plugin and fails to produce a working .dylib for the package. This means the plugin is not loaded when MITK-GEM is started. Happens on OSX only.

It appears that the problem is related to the dependency to org.mitk.gui.qt.common.legacy which includes Qt SQL support. On OSX (and thus Qt4), this ends up trying to include PostgreSQL which for some reason does not work. PostgreSQL is actually not needed at all, there is 0 dependency on that in all of MITK. The issue would probably be resolved by moving to Qt5.

For now, i don't see an easy solution for this. Qt5 support for OSX is hopefully included in the next MITK release.

It can take quite long to complete those actions and some kind of visual feedback would be nice.

I followed the instructions on http://araex.github.io/mitk-gem-site/ (FAQ section) for using the CLI.
If I got it right, I have to run a file named "GraphCutSegmentation", however I cannot find it in the downloaded folder after extraction. Is there anything else I have to install, or do I have to build from source?

Loading a .matmap file crashes on windows. Saving works fine.

I want to use your software to segment the liver. How do I set the following three parameters: sigma, boundary direction, foreground label value?

In order to build mitk-gem, I need to build MITK-mitk-gem from https://github.com/araex/MITK.git,which include CGAL configuration. But I met some problems in building this version of MITK.

I've just wrapped gridCut which is really faster and much more memory effcient. The problem is I don't know how to manage the licensing part. If you have any idea on the way to legally plug it into mitk-gem, I would be glad to pull request this new feature.
Otherwise, if we find a way, we could even imagine to implement the multi-label graphCut using the new features of MITK 2016.11 & the multi-label graphCut support of gridCut.

1. Load Nifti (nii) image with given affine matrix as stored with the image.
2. Click Segmentation tab.
3. Add new segmentation, specify name.
4. The image changes orientation and is now upside down in the axial window.

The power law widgets were originally a separate group and not nested inside the options group. This was introduced during the qt5 specific fixes by accident. However, i'm not sure whether it's better this way?

Wait for feedback for alpha1 and fix if needed.

hello,I'm very confused by how to deploy MITK-app ?
I had build My MITK-application （debug is ok,but release occur to problem:QWidget:Must construct a QApplication before a QWidget) like GEM ,but I don't know how to make .exe or .zip file，can someone give me a suggestion?Thanks!
I can't find out above problem .

What is the abstract idea / intention of the extend image step that can be configured via the material mapping parameters?

This is what happens: each extend step builds a weighted average of the voxels just outside of the mesh stencil (think a shell around the mask). Why is this done, what does this accomplish and why would one do this more than once?

Relates to issue #3

I received a dataset in the .mitk format ("save project"). I'm not sure whats going on, but when i load it via "load project" it just keeps crashing at random steps during the material mapping. When i load the content of the .mitk manually, everything works fine.

Might have something to do with data node properties being loaded with the .mitk file?

I have a problem when I build this project in windows 10, there is a problem about CGAL. I have installed CGAL in my computer, but it still come out a problem said lack of package CGAL. How to solve this problem.

I've been trying to build the master branch but it requires qtWebEngine. I couldn't find this library packaged for ubuntu 16.04 and I get weird errors when I try to build WebEngine : no matching function for call to ‘QMenu::addAction(QString, QApplication*, void (&)()).
Then, the question is, how did you solve the WebEngine dependency for linux ? Or could you build a beta 2016.3 release ?

I don't have any explanation to this but the the mesher returns inverted faces surfaces.

If the node ids in a .vtu are not sequential or do not start at ID 1, MITK will get stuck in an allocation loop for the geometric primitives. Eventually this will crash the application with a bad allocation / out of memory exception.

The following code:

padder->SetInputData(_img);
voi->SetInputConnection(padder->GetOutputPort());
voi->Update();

changes the scalar type:

(lldb) p _img->GetScalarTypeAsString()
(const char *) $6 = 0x0000000117764473 "float"
(lldb) p padder->GetOutput()->GetScalarTypeAsString()
(const char *) $3 = 0x000000011776446c "float"
(lldb) p voi->GetOutput()->GetScalarTypeAsString()
(const char *) $4 = 0x0000000117764473 "double"

This is unexpected because vtkExtractVOI is supposed to request its output information from the padder.
The only thing i can find is a 12 year old bug (that is still open) in paraview: http://www.paraview.org/Bug/print_bug_page.php?bug_id=380 . One would assume this is something that would pop up more often, not sure if i'm just overlooking something.

The easiest fix would be to just add a cast or work with doubles exclusively. I'll look into it.

Hi,
Since I am not an experienced programmer, I wanted to ask, if there is a relatively easy way to automate the material mapping process? (1. uploading the CT-images (DICOM) and the corresponding volumetric meshes 2. material mapping 3. exporting the output files)

Thanks in advance

Hello, I found the imagecropper plugin in the binary file, but it is not found in the source code. Is this part not open?

The current material mapping results are slightly different from the standalone ones. This is because we've yet to replicate the original behavior on points near the image boundary.

Approaches:

• Pad with wrap around. Implemented in 2e5d29
• Pad with constant value.
• Pad with nearest neighbor (i'm rather certain that this is what happened in the standalone version.)
• Clamp the indices and let the interpolation filter handle out of bounds points

Hi,
I am currently trying to perform 3d modeling of femur based on CT scan data. However I find myself struggling with the process in material mapping module. I've followed the video in tutorial page to generate model from my CT scan data, including all the steps and equation used in material module, but after i opened my txt material file, the resulting young modulus in the material mapped mesh text document turns out to be zero for all nodes and elements. When i tried used CT scan dataset provided in the tutorial page, it works properly.

Does it has something to do with my CT scan settings? I used 499mA and 100 kV.
Or are there known issues or considerations when working with higher mA settings in CT data that could lead to zero values in material mapping?

Thanks in advance.

Currently we only have the 4 options that where set in the standalone program. Do we need some more configuration?

Annoying since a global reinit also looses resets the currently viewed slice. The reinit is probably necessary to make sure the VTK interactor works correctly, but maybe there's a way to restore the currently selected slice?

The volume mesher (and tetgen) needs a closed surface in order to produce a valid output. Providing an open surface will result in an empty output. Since we do not render the volume meshes, the user might not even notice the error and just proceed. At no point in the pipeline will this cause any errors, but it should at least trigger a warning to inform the user that something isn't right.

Hello,
I have gotten a good result of bone segmentation by using your software, I do not know whether it can be used in liver segmentation.

Not sure if it is actually a tetrahedral mesh saving problem - it is.

@araex found that the 'Save Project' is called SceneIO at the MITK level and write for unstructured grids is missing.

Submitted on Jan 3, 2017:
'I have some issues with the MITK-GEM software. If I want to store what I have worked on, it does not store the tetrahedral mesh. I attached the error message of MITK-GEM. Do you know what the problem is?'

Update on Jan 4, 2017:
Same happens on MacOSX:

INFO|Wed Jan 4 08:24:42 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/SceneSerialization/src/mitkSceneIO.cpp(e2)|SaveScene|0|MitkSceneSerialization|Storing scene with 6 objects to /Users/pauc/Desktop/test-image-surface-tetmesh.mitk
INFO|Wed Jan 4 08:24:42 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/Core/src/IO/mitkItkImageIO.cpp(173)|Write|0|MitkCore|Writing image: /var/folders/gg/97_t00fn7l967t16dxkyv5c40000gn/T/SceneIOTempUID_20170104152442182904/aaaaaa_input.nrrd
INFO|Wed Jan 4 08:24:43 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/Core/src/IO/mitkItkImageIO.cpp(173)|Write|0|MitkCore|Writing image: /var/folders/gg/97_t00fn7l967t16dxkyv5c40000gn/T/SceneIOTempUID_20170104152442182904/baaaaa_input_graphcut_picked.nrrd
INFO|Wed Jan 4 08:24:43 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/Core/src/IO/mitkItkImageIO.cpp(173)|Write|0|MitkCore|Writing image: /var/folders/gg/97_t00fn7l967t16dxkyv5c40000gn/T/SceneIOTempUID_20170104152442182904/daaaaa_input_graphcut.nrrd
ERROR|Wed Jan 4 08:24:43 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/SceneSerialization/src/mitkSceneIO.cpp(1c3)|SaveBaseData|0|MitkSceneSerialization|No serializer found for UnstructuredGrid. Skipping object
ERROR|Wed Jan 4 08:24:43 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/SceneSerializationBase/src/mitkPropertyListSerializer.cpp(7a)|SerializeOneProperty|0|MitkSceneSerializationBase|No serializer found for GridRepresentationProperty. Skipping object
ERROR|Wed Jan 4 08:24:43 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Modules/SceneSerializationBase/src/mitkPropertyListSerializer.cpp(7a)|SerializeOneProperty|0|MitkSceneSerializationBase|No serializer found for GridVolumeMapperProperty. Skipping object
WARN|Wed Jan 4 08:24:44 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Plugins/org.mitk.gui.qt.ext/src/QmitkExtFileSaveProjectAction.cpp(9d)|Run|0|org_mitk_gui_qt_ext|The following nodes could not be serialized:

• UnstructuredGrid contained in node 'tetrahedral mesh'
  WARN|Wed Jan 4 08:24:44 2017 ||/Users/admin/TC/buildAgent/work/1b4b341ba65bea09/build/MITK/Plugins/org.mitk.gui.qt.ext/src/QmitkExtFileSaveProjectAction.cpp(ad)|Run|0|org_mitk_gui_qt_ext|The following properties could not be serialized:
• GridRepresentationProperty associated to key 'grid representation'
• GridVolumeMapperProperty associated to key 'volumerendering.mapper'

As requested by Ben.

The command line options are not explained. Add descriptions.

It's possible to enter power laws that will result in domain errors during the std::pow call [1]. Specifically, if the base is finite and negative and the exponent is finit and non-integer (=> rho smaller 0) it will result in a domain error and return NaN.

Usually our meshes should always have positive rho values, but we should still consider how to handle this as i'm not sure how the later steps (FE solver) handles NaNs. I think we should either prevent the user from entering power laws for negative value ranges (effectively enforce minimum element value >= 0) or somehow handle it gracefully.

Thoughts?

[1] http://en.cppreference.com/w/cpp/numeric/math/pow

(Reported by Ben): When I switch off the erosion step I get an error in the output (see attached). It is almost like half of the stack is missing. I first thought this was due to my mesh but I tried mapping using Halldor’s vtu file of the same bone and I go a similar result.

This would allow to:

• Change version info in title bar
• Add a custom icon to the titlebar
• Set predefined perspectives (opened plugins, size of bars)
• Show / hide plugins from the hotkey bar

I was thinking about starting developing a multi label graphCut (using gridCut library) feature soon. Problem is, I don't know how to manage the GUI part. Do you have any idea how could we manage a user-friendly selection of an arbitrary number of labels in MITK ?
Thanks.

Hello,
Following the FEZ workshop I wanted to ask to the MITK-GEM developers if there is any future release of the software planned. I am asking this in the perspective of adopting the software for our finite element workflows and student projects: the software and its features look great but without future public updates or releases it is difficult to plan a long-term use.
Thank you,
Luca

It seams like Qt5.5 is no longer available in https://download.qt.io/archive/qt/. Compiling MITK-GEM on Ubuntu 20.04 is giving me troubles when I try another version of Qt, especially with: "Qt5WebKitWidgets" provides a separate development package". Is there a solutions for this problem?

Hi, I am currently trying to perform Ct-calibration using a K2HPO4-Phantom and I am struggeling to find the relation between ρ_ct density mg dipotassium phosphate per cc and mg hydroxyapatite per cc . Does anyone know how to convert mg dipotassium phosphate per cc to mg hydroxyapatite per cc?

Not sure what pattern it follows, but it's changing quite often. I'm pretty sure it's an issue of the segmentation plugin not obeying the global preference.