View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0003779Slicer4Module EMSegmentpublic2014-07-27 11:502018-03-02 11:06
Reporterpohl Assigned Tohjmjohnson  
PriorityhighSeveritymajorReproducibilityalways
Status closedResolutionfixed 
Platformall platformsOSOS Version
Product Version 
Target VersionSlicer 4.4.0Fixed in VersionSlicer 4.4.0 
Summary0003779: 'itk::ExceptionObject exception is thrown in TransformFileWriterTemplate which is run in connection with BRAINS
Description

Steve: I looked a bit into one of the failing tests EMSegCL_Task_MRIHumanBrain_2CH [1] which is failing on all platforms.

The test results are hard to read on cdash, but if I put it in an editor I can format it to the attached version.

The key error message is about not being able to crate the transform file writer:

=============================== ITKv4 Registration: Starting Transform Estimations for BSpline(4 of 4).===============================

PRE : 0 2187
POST: 2187 2187
PRE : 0 2187
POST: 2187 2187
Moving image is warped by initial transform, before it is passed to the BSpline registration.
Running bspline registration (meshSizeAtBaseLevel = [6, 6, 6])

WARNING: LBFGSB optimizer does not support scaling. All scales are set to one.
Set the initial position of the optimizer:
Stop condition from LBFGSBoptimizer.LBFGSBOptimizerv4: Function tolerance reached after 47 iterations. The relative reduction of the cost function <= 2.22045e-07 = CostFunctionConvergenceFactor (1e+09) * machine precision (2.22045e-16).

Write the output composite transform to the disk ...
terminate called after throwing an instance of 'itk::ExceptionObject'
what(): /software/Slicer4/superbuild/ITKv4/Modules/IO/TransformBase/include/itkTransformFileWriter.hxx:200:
itk::ERROR: TransformFileWriterTemplate(0x2040630): Can't Create IO object for file /tmp/EMS9027344003921ht_vtkMRMLBSplineTransformNode1Composite.h5')

604: (<type 'exceptions.SyntaxError'>, SyntaxError('invalid syntax', ('<string>', 1, 5691, "__tpycl_result = slicer.IIJnfVMh.PrintText('TCL: HACK: NumberOf Threads -1

Original Fixed image origin[119.531, 92.25, 119.531, 0]

WARNING --numberOfSamples is deprecated, please use --samplingPercentage instead
WARNING: Replacing command line --samplingPercentage 0.00483092\nTranformTypes: Rigid(1 of 4).

TranformTypes: ScaleSkewVersor3D(2 of 4).

TranformTypes: Affine(3 of 4).
TranformTypes: BSpline(4 of 4).

I suspect many of the emsegment tests fail for similar reasons. I didn't find much about the TransformFileWriterTemplate and slicer except some back-and-forth when it was first introduced [2].

[1] http://slicer.cdash.org/testDetails.php?test=5573281&amp;build=251384

[2] http://www.itk.org/pipermail/insight-developers/2013-July/022956.html

Steps To Reproduce

ctest -VV -R EMSegCL_Task_MRIHumanBrain_2CH

TagsNo tags attached.

Relationships

related to 0003561 closedpohl Segmentation Task "Hand Bone" crashes 
related to 0002741 closedpohl Fix EMSegCLTask{HumanEye, MRIHumanBrainExp}_BRAINS_small failing on windows 

Activities

2014-07-27 12:31

 

blub.log (45,829 bytes)
pohl

pohl

2014-07-28 15:04

developer   ~0012262

Here is a faster and easier way to replicate the error

Step 1: Download

http://svn.slicer.org/Slicer3/branches/Slicer4-EMSegment/Testing/TestData/MRI-Human-Brain/very_small_MRIHumanBrain_T1_aligned.nrrd

and

http://svn.slicer.org/Slicer3/branches/Slicer4-EMSegment/Testing/TestData/MRI-Human-Brain/very_small_atlas_t1.nrrd

Step 2 Execute

/software/Slicer4/superbuild/Slicer-build/lib/Slicer-4.3/cli-modules/BRAINSFit --backgroundFillValue 0.0 --interpolationMode Linear --outputVolumePixelType short --numberOfSamples 100 --numberOfIterations 3 --useRigid --useScaleSkewVersor3D --useAffine --useBSpline --splineGridSize 6,6,6 --maxBSplineDisplacement 0 --useCachingOfBSplineWeightsMode ON --initializeTransformMode useCenterOfHeadAlign --minimumStepLength 0.005 --translationScale 1000 --reproportionScale 1 --skewScale 1 --maskProcessingMode NOMASK --numberOfHistogramBins 40 --numberOfMatchPoints 10 --costMetric MMI --fixedVolumeTimeIndex 0 --movingVolumeTimeIndex 0 --numberOfThreads -1 --debugLevel 0 --failureExitCode -1 --fixedVolume very_small_MRIHumanBrain_T1_aligned.nrrd --movingVolume very_small_atlas_t1.nrrd --outputVolume /tmp/BRAINS_OUT.nrrd --bsplineTransform /tmp/BRAINS_OUT.mat
pieper

pieper

2014-07-28 15:42

administrator   ~0012263

I can reproduce what Kilian reports. Running BRAINSFit with the data he provided results in the following output.

WARNING: LBFGSB optimizer does not support scaling. All scales are set to one.
Set the initial position of the optimizer:
Stop condition from LBFGSBoptimizer.LBFGSBOptimizerv4: Function tolerance reached after 22 iterations. The relative reduction of the cost function <= 2.22045e-07 = CostFunctionConvergenceFactor (1e+09) * machine precision (2.22045e-16).
Write the output composite transform to the disk ...
libc++abi.dylib: terminating with uncaught exception of type itk::ExceptionObject: /Users/pieper/slicer4/latest/Slicer-superbuild/ITKv4/Modules/IO/TransformBase/include/itkTransformFileWriter.hxx:200:
itk::ERROR: TransformFileWriterTemplate(0x7fb373100070): Can't Create IO object for file /tmp/BRAINS_OUTComposite.h5
Abort trap: 6
pieper

pieper

2014-07-31 07:28

administrator   ~0012316

The missing transform IO seems to be a packaging issue.

When I use my local build, brainsfit can write out a .h5 file. But when using the Nightly build it cannot.

I tested Slicer-4.3.1-2014-07-28-macosx-amd64, the most recent on the download page as of today.
pieper

pieper

2014-07-31 07:49

administrator   ~0012318

Scratch the last note - It's an older package (no new mac package last night)
jcfr

jcfr

2014-09-02 11:02

administrator   ~0012453

Fixed in r23638 / r23640

See http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&amp;revision=23638
See http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&amp;revision=23640

Issue History

Date Modified Username Field Change
2014-07-27 11:50 pohl New Issue
2014-07-27 11:50 pohl Status new => assigned
2014-07-27 11:50 pohl Assigned To => pohl
2014-07-27 12:31 pohl File Added: blub.log
2014-07-27 12:39 pohl Description Updated
2014-07-28 15:04 pohl Note Added: 0012262
2014-07-28 15:42 pieper Note Added: 0012263
2014-07-29 09:00 pohl Assigned To pohl => hjmjohnson
2014-07-29 16:31 jcfr Relationship added related to 0002741
2014-07-29 16:31 jcfr Relationship added related to 0003561
2014-07-31 07:28 pieper Note Added: 0012316
2014-07-31 07:49 pieper Note Added: 0012318
2014-09-02 11:02 jcfr Note Added: 0012453
2014-09-02 11:02 jcfr Status assigned => resolved
2014-09-02 11:02 jcfr Fixed in Version => Slicer 4.4.0
2014-09-02 11:02 jcfr Resolution open => fixed
2018-03-02 11:06 jcfr Status resolved => closed