[Insight-developers] Very suspicious behavior.

[Luis Ibanez]

Hi Hans,

This is not quite a bug.

It is intended behavior based on the fact that the RegistrationMethod,
not the Metric, is the one that has a Filter API.

If you want to control the number of threads, you can put the setting
in the registration method.

                       registration->SetNumberOfThreads( 1 );

Just as you would do with any multi-threaded filter.


    Luis


---------------------------------------------------
On Mon, Feb 28, 2011 at 8:09 AM, Johnson, Hans J <hans-johnson at uiowa.edu> wrote:
> Dan,
> I found a tangentially related bug in ITKv3 & ITKv4 that was complicating my
> understanding of the BRAINSFit threading issues that you are so kindly
> looking into.
> Rather than make the entire program single threaded, I wanted to isolate and
> force just itk::MattesMutualInformationImageToImageMetric  into single
> threaded mode:
> BRAINSCommonLib/BRAINSFitHelper.cxx :  Inserted line 115:
> 106   if(this->m_CostMetric == "MMI")
> 107     {
> 108     typedef
> itk::MattesMutualInformationImageToImageMetric<FixedVolumeType,MovingVolumeType>
> MetricType;
> 109     this->SetupRegistration<MetricType>();
> 110
> 111     MetricType::Pointer localCostMetric =
> this->GetCostMetric<MetricType>();
> 112
> localCostMetric->SetNumberOfHistogramBins(this->m_NumberOfHistogramBins);
> 113     const bool UseCachingOfBSplineWeights = (
> m_UseCachingOfBSplineWeightsMode == "ON" ) ? true : false;
> 114
> localCostMetric->SetUseCachingOfBSplineWeights(UseCachingOfBSplineWeights);
> 115     localCostMetric->SetNumberOfThreads(1);
> 116     this->RunRegistration<MetricType>();
> 117     }
> The problem was the that ImageRegistrationMethod was modifying the state of
> my localCostMetric and resetting the number of threads to a larger value.
> I've posted patches to both ITKv3 and ITKv4 to avoid changing the
> prescribed localCostMetric number of threads:
> http://review.source.kitware.com/#change,1088
> http://review.source.kitware.com/#change,1087
> =============================================
> Once the ITKv[3,4] bugs are fixed, then I can sacrifice time to get a
> consistent result in BRAINSFit ( by forcing localCostMetric for MMI to  only
> use 1 thread).
> =============================================
> I hope you can find a way to give consistent threaded results from "MMI"
> because that greatly increases the processing time when it is single
> threaded.
> Thanks,
> Hans
>
>
> ________________________________
> Notice: This UI Health Care e-mail (including attachments) is covered by the
> Electronic Communications Privacy Act, 18 U.S.C. 2510-2521, is confidential
> and may be legally privileged.  If you are not the intended recipient, you
> are hereby notified that any retention, dissemination, distribution, or
> copying of this communication is strictly prohibited.  Please reply to the
> sender that you have received the message in error, then delete it.  Thank
> you.
> ________________________________
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://kitware.com/products/protraining.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-developers
>
>