[Insight-users] Problem with itkN3MRIBiasFieldCorrectionImageFilter

Wed Oct 14 20:44:04 EDT 2009

Well, that's good. Let me know what you find out.

On Oct 14, 2009 8:17pm, Ricardo Ferrari <ricardo.jose.ferrari at gmail.com>  
wrote:
> Hi Nick,

> The most funny thing!!! The error I was getting was when using my  
> PC-workstation at University. Now, I am at home and I have just checked  
> out the code from my svn repository (the same one I was testing at Univ)  
> and I am not getting the error anymore.

> I believe it is something related to the ITK lib. I have re-compiled the  
> svn ITK in my home machine last night.

> Tomorrow, I am going to compare the ITK configurations to see if there is  
> any difference.

> I will keep you update on this.

> Thanks again,
> Ricardo

> On Wed, Oct 14, 2009 at 9:01 PM, Nicholas Tustison ntustison at gmail.com>  
> wrote:

> Hi Ricardo,

> Okay, now I understand. In that case, I have two other suggestions:

> 1) In the output of your code below

> cout GetOrigin() cout GetSpacing()
> cout GetLargestPossibleRegion().GetSize() cout GetDirection() cout  
> m_NumberOfFittingLevels
> cout m_SplineOrder cout m_NumberOfControlPoints

> can you also add

> cout GetNumberOfPoints()

> 2) Also, could you see what happens if you convert your minc file to  
> nifti or some other format?

> Nick

> On Oct 14, 2009, at 7:55 PM, Ricardo Ferrari wrote:

> Hi Nick,

> Thanks for your e-mail.

> Actually, I am not submitting a mask of all zeros because I am saving and  
> checking the resulting mask image obtained from the Otsu threshold.

> What I´ve meant by "m_maskImage is purposely equal to zero" in the source  
> code is that I have set the pointer to 0 (NULL) so that the mask could be  
> computed by using the Otsu thresh. As I said, the mask seems to be okay -  
> I have visually checked it.

> The funny thing is if I do not downsample the image (and the mask, of  
> course) I have no problems. I have also checked both the downsampled  
> image & mask, and apparentely (by visually checking) everything is fine.

> I will keep trying to figure out the problem!!! Any suggestion is  
> welcome !!

> Thanks,
> Ricardo

> On Wed, Oct 14, 2009 at 7:29 PM, Nicholas Tustison ntustison at gmail.com>  
> wrote:

> Hi Ricardo,

> If you're submitting a mask of all zeros, that would explain the error  
> message you're getting. Instead of submitting a mask of all zeros, just  
> type a non-existent filename and the otsu mask will be calculated. Let me  
> know if that works out.

> Nick

> On Oct 14, 2009, at 5:10 PM, Ricardo Ferrari wrote:

> Dear itk users,

> I am trying to run the itk N3 algorithm implemented by Nicholas J.  
> Tustison and James C. Gee and I am getting a "segmentation fault" error.

> Above I am sending the main part of my code and the error I am getting.

> -------------------------------------
> /// m_inputImage is a minc image - you can download it from the following  
> site: http://www.4shared.com/file/140891877/c11d6f0c/20090808_0_DUAL.html

> int m_shrinkFactor = 2;

> /// m_maskImage is purposely equal to zero so that it can be computed  
> using Otsu threshold

> ///
> /// Use Otsu threshold to compute a simple mask, if one is not provided

> ///
> if( ! m_maskImage )
> {
> typedef itk::OtsuThresholdImageFilter ThresholderType;
> typename ThresholderType::Pointer otsu = ThresholderType::New();

> otsu->SetInput( m_inputImage );
> otsu->SetNumberOfHistogramBins( 200 );
> otsu->SetInsideValue( 0 );
> otsu->SetOutsideValue( 1 );
> otsu->Update();

> m_maskImage = otsu->GetOutput();

> }

> ///
> /// DownSampling the input image to reduce computation time
> ///
> typedef itk::ShrinkImageFilter ShrinkerImageType;
> typename ShrinkerImageType::Pointer shrinker1 = ShrinkerImageType::New();

> shrinker1->SetInput( m_inputImage );
> shrinker1->SetShrinkFactors( 1 );

> typedef itk::ShrinkImageFilter ShrinkerMaskType;
> typename ShrinkerMaskType::Pointer shrinker2 = ShrinkerMaskType::New();

> shrinker2->SetInput( m_maskImage );
> shrinker2->SetShrinkFactors( 1 );

> if( m_shrinkFactor > 1 )
> {
> shrinker1->SetShrinkFactors( m_shrinkFactor );
> shrinker2->SetShrinkFactors( m_shrinkFactor );

> shrinker1->Update();
> shrinker2->Update();
> }

> typename TRealImageType::Pointer realInputImage = shrinker1->GetOutput();
> typename TMaskImageType::Pointer maskShrinked = shrinker2->GetOutput();

> ///
> /// Run bias field correction
> ///
> typedef itk::N3MRIBiasFieldCorrectionImageFilter CorrectorType;
> typename CorrectorType::Pointer corrector = CorrectorType::New();

> corrector->SetInput( realInputImage );
> corrector->SetMaskImage( maskShrinked );
> corrector->SetMaskLabel( NumericTraits::One );
> corrector->SetMaximumNumberOfIterations( m_maxNumIter );

> corrector->SetNumberOfFittingLevels( m_numFittingLevels );

> try
> {
> corrector->Update();
> }
> catch ( itk::ExceptionObject & err )
> {
> std::cout

> std::cout throw err;
> }
> catch (... )
> {
> std::cout

> throw;
> }

> OUTPUT
> --------------------------
> Starting
> [-116.98, -103.715, -34.2648]
> [0.857143, 0.857143, 6]
> [280, 280, 25]
> 0.999208 -0.00996255 -0.0385217
> 0.0108005 0.999708 0.0216062

> 0.0382952 -0.0220051 0.999024

> [3, 3, 3]
> 3
> [4, 4, 4]
> Segmentation fault
> --------------------------

> After spending some time looking for the problem I found out that the it  
> is happening in the following part of the  
> itkN3MRIBiasFieldCorrectionImageFilter.txx file. The problem does not  
> occur if I use the original image without downsampling it.

> I really appreciate any help on this !!!

> Thanks,
> Ricardo

> cout
> cout GetOrigin() cout GetSpacing()

> cout GetLargestPossibleRegion().GetSize() cout GetDirection() cout  
> m_NumberOfFittingLevels

> cout m_SplineOrder cout m_NumberOfControlPoints
> typename BSplineFilterType::Pointer bspliner = BSplineFilterType::New();
> bspliner->SetOrigin( fieldEstimate->GetOrigin() );

> bspliner->SetSpacing( fieldEstimate->GetSpacing() );
> bspliner->SetSize( fieldEstimate->GetLargestPossibleRegion().GetSize() );
> bspliner->SetDirection( fieldEstimate->GetDirection() );
> bspliner->SetGenerateOutputImage( true );

> bspliner->SetNumberOfLevels( this->m_NumberOfFittingLevels );
> bspliner->SetSplineOrder( this->m_SplineOrder );
> bspliner->SetNumberOfControlPoints( this->m_NumberOfControlPoints );
> bspliner->SetInput( fieldPoints );

> bspliner->Update();

> cout

> _____________________________________
> Powered by www.kitware.com

> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html

> Please keep messages on-topic and check the ITK FAQ at:  
> http://www.itk.org/Wiki/ITK_FAQ

> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20091015/8e48e77c/attachment-0001.htm>