Thx Marco. Works for me with Rev. 20563 installer.
Will remove the Debug output, then close.
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
Aug 2 2016
Removed debug output -> Resolved.
[SVN revision 20375]
FIX (#3050): Quickfix: Abort button now only stops segmentation; model remains in undefined state and is not removed. But does not crash app any more.
[SVN revision 20569]
FIX (#3049): Removed debug output; (Also made some error output more presentable).
[SVN revision 20377]
FIX (#3050): Controlled abort of segmentation. A binary image is still created. Abort can also be used to stop the segmentation when quality is sufficient.
As a quick fix, you can copy the file
/home/kast/public-html/Lymphnode_SM.zip
and extract it in your
/home/.mitk folder.
Cannot work in the moment with the installer.
The lymph node shape and appearance model are not distributed with MITK.
See T2979 for more information.
[SVN revision 20368]
FIX (#3049): Use binary version of appearance model
Seems to work now.
[SVN revision 20359]
FIX (#3003): Try to reset the ShapeModelGeometry3D when cleaning ShapeModelData
[SVN revision 20322]
FIX (#3003): Clean up Shape Model Data after usage. Mostly fixes this problem.
[SVN revision 20317]
FIX (#3003): Make ClearData method public
[SVN revision 20313]
FIX (#3003): Add ClearData method for ShapeModelData
Shape model should now be able to segment a single time step of a 4D image.
Time step has still to be set 'outside' of segmentation perspective, but segmentation is done on the correct time step, as is the creation and display of binary image.
Not necessary, seems to work as expected.
[SVN revision 20213]
FIX (#2988): 4D image support for shape models in Segmentation module; also: Removed memory leak when creating binary image; adapted names for load to binary shape models
Forgot adding keyword 3M3
[SVN revision 20360]
FIX (#2987): Also clear the model nodes
Pressing the "Abort Segmentation" leads to the same (similar?) problem.
Considering this critical.
Missing step:
Positively tested with current installer -> Works.
But, actual bug still not fixed.
[SVN revision 20361]
FIX (#2987): Do not clean filter on startup; set aborted to false again after abortion. Only clean filter once.
Bug still not fixed. But works for me now in Release mode.
Two ideas where it could come from:
- ShapeModelData is not cleared correctly. A Problem, but most likely not cause of the bug.
- Something with the tool event sending/receiving or methods triggered by them.
Problem does not occur anymore when commenting out both:
m_ToolManager->NewNodesGenerated.Send(); (ln 249)
m_ToolManager->ActivateTool( -1 ); (ln 251)
in ShapeModelTool.
Does no longer block #3049, as the models are distributed with the installer.
Verified, works for me with installer Rev. 20568.
Does no longer block #2698, then, as models are theoretically always available. Will open a new bug for CD distribution.
Special case is the Lymph node shape model: 2 Files, together 72 kB.
These are in the moment not supported to be loaded from the internet.
Could these be somehow included in the installer? E.g. as Qt resource?
[SVN revision 20596]
FIX (#2973): Signal was not emitted any more when tool was zero.
[SVN revision 20312]
FIX (#2973): Quickfix: Only emit button clicked signal when not a shape model tool.
Reopened.
Sometimes slice interactor is not restored after lymph node segmentation finished.
Known issue: See T2987
It is still possible to press continue after crash.
-> Works for me.
Would it be possible to not show the model at all in data manager, then?
At least for 3M, the user should anyways not be able to do something with the model except for using it for segmentation.
[SVN revision 20247]
FIX (#2929): Use binary appearance model
[SVN revision 20235]
FIX (#2928): Don't use matrix for LV wall models because of shear strain
Have tried some matrix initialization from the image.
Still initializes shape model upside down for images with non-CT geometry.
As debugging is not possible due to non-stop exceptionsin the "segmentation" module, I am not able and willing to continue working on this bug.
I will not remodel support for non-uniform geometries again.
Changed for all models;
Hannes, added a zip with lymph node models to your directory.
Would it be possible to load them via internet, too?
[SVN revision 20184]
FIX (#2928): Fix liver default init (actually it never worked before...)
Got all except liver model. Will convert it.
[SVN revision 20248]
FIX (#2929): Use binary shape model
Okay, fixed.
For the LV wall models, the geometry matrix is a matrix product of rotation and shear strain. Because the models were trained on the sheared data, they do not need to be sheared again explicitely.
But, sole rotation can't be extracted or estimated from the matrix.
[SVN revision 20092]
FIX (#2928): Initialize shape model by image matrix
Changed title; password protection should not be necessary.
[SVN revision 20174]
FIX (#2928): Only do default transform and matrix init if no automatic method is available
[SVN revision 21076]
FIX (#2927): Make 2D index coordinates corner based.
AbstractTransformGeometry already tests this more or less -> All tests added -> Closed.
Trying to fix this today.
-> Requesting core mod again.
[SVN revision 21199]
ADD (#2927): Simple test for coordinate system consistency
Added change request. Requesting core mod.
I talked about this with Daniel S. yesterday. Don't think it is a coordinate system problem.
Rendering problems without initialization are quite obvious, from where should the renderer, SliceNavigationController etc. now the image geometries?
Problem does not occur any more.
Most likely because images are now reinited after loading.
Note: This does not actually solve the initial problem, which is still unknown.
Not really 3M3 relevant.
Most likely not so great effect on execution speed.
[SVN revision 20007]
FIX (#2401): Adapt convenience method for WorldToIndex with Itk Index to
corrected MITK coordinate system, and therefore rounding.
Fixed by using ->GetInverse.
Relocated Multiplication with inverse matrix to seperate method
MultiplyWithWorldToIndex.
It seems that in this function the 0.5 pixel offset is calculated a second time (but wrong for negative values(?) ).
Req. core mod.
Thomas or Daniel,
can you try to run a long lasting registration with and without these lines:
Req. core mod
[SVN revision 20058]
FIX (#2884): Use stable matrix inversion; relocated multiplication to separate function.
Fixed by above commit
Looking at mitk::Geometry3D, there are still some remnants from before voxel center change, as well as not adapted mehtods (not noticed because they are not used). This class has to be cleaned up for the new coordinate system.
Not the pixel offset is changed, but rounded.
Anyhow, the rounding is not compatible with the changed coordinate systems anymore.
Must use floor().
Patch for possible removal of NaN check in inner loop
Added patch. Daniel, please have a look at it.
Patch for possible removal of NaN check in inner loop
[SVN revision 19959]
FIX (#2874): Add SetPoints() method to mitkPointLocator which takes itk PointSets
[SVN revision 19964]
FIX (#2877): Removed options MITK_ITK_SOURCE_DIR and MITK_VTK_SOURCE_DIR in CMake GUI
Closed.
Closing anyway to keep bug list more managable.
Test T2442 is still open.
Due to ITK templatings, PointSet has to be further defined...
[SVN revision 19987]
FIX (#2874): Further define itk::PointSet template. Also use check if rebuild needed for Mitk and Itk point sets.
Not 3M3 relevant.
Okay, not closed because tests are missing. (#2442).
But fixed.
Fixed -> Closed.
Fixed.
Will also add checks if MITK or ITK pointsets have changed for speed considerations. (As already done for Vtk pointsets.)
Just found it:
Warning: org.mitk.gui.qt.basicimageprocessing is missing requirements and won't be built. Missing: MitkSB;MitkDiffusionImagingMBI
I will check it.
Seems to work.
Let's check tomorrows dart client, then close.