[Insight-users] register of MRI and a partial fMRI

[Yu Qi]

Hi Karthik,
 
Thanks!
 
I found the problem: there was something wrong with my region.
It is OK now.
But by using SetFixedImageRegion(), I didn't see much difference in running time.
 
And I am not quite sure about "initial transform". Do you mean I should set a initial transform that is close to the registration result?
What is the function of the code below?
> VersorType rotation;
> VectorType axis;
>
> axis[0] = 0.0;
> axis[1] = 0.0;
> axis[2] = 1.0;/
> / const double angle = 0;/
> / rotation.Set( axis, angle );/
> / transform->SetRotation( rotation );/

I use only 10k samples in registration and cost around 5 min, is it normal?
 
Thanks!
 
Best regards,
Yu Qi

Karthik Krishnan <Karthik.Krishnan at kitware.com> wrote:
Hi Yu,

I think the problem is that you are using SetRegion(). Instead you
should be using SetFixedImageRegion(). You do not want to modify the
co-ords of the image, rather restrict computation of the metric to the
region where the brain is (+ a border as suggested below). This speeds
up registration and also ensures that the registration is not affected
by irrelevant data.

Please print out the initial transform. If you say that it works well
before setting the region, your initial transform should be close to the
one you got prior to setting the regions.

thanks
karthik

Yu Qi wrote:

> Hi Karthik,
> I did use the initializer to set centre. But I wonder if it matters
> that I set region before set centre. Does one influence the other?
> The tranform parameters used in my application are copied from example
> ImageRegistration8.cxx as follows
> /VersorType rotation;
> VectorType axis;
>
> axis[0] = 0.0;
> axis[1] = 0.0;
> axis[2] = 1.0;/
> / const double angle = 0;/
> / rotation.Set( axis, angle );/
> / transform->SetRotation( rotation );/
>
> and the/ translationScale = 1.0 / 1000.0;/
>
> / optimizer->SetMaximumStepLength( 0.2000 );
> optimizer->SetMinimumStepLength( 0.0001 );/
>
> my data sets are MRI 256X256X256
>
> and fMRi 64X64X28
>
> is there any problems with these settings?
>
> Thank you!
>
> Best regards,
>
> Yu Qi
>
>
>
> */Karthik Krishnan /* wrote:
>
> Hi Yu,
>
> Print out the parameters of the transform. There just isn't enough
> overlap between the fixed and the moving image to start with (or the
> transform is driving you too far away too soon). See
> itkMattesMutualInformationImageToImageMetric.txx
>
>
> Line: 702
> if( nSamples < m_NumberOfSpatialSamples / 4 )
> {
> itkExceptionMacro( "Too many samples map outside moving image
> buffer: "
> << nSamples << " / " << m_NumberOfSpatialSamples << std::endl );
> }
>
>
> Remember that if you are using the centered transform, set the center
> (for example with the CenteredTransformInitializer helper class).
>
> See Examples/Registration/ImageRegistration9.cxx
>
> Thanks
> kk
>
> Yu Qi wrote:
>
> > Hi Luis,
> > Thanks for your instruction.
> > I tried to set region to the fixed image by setting
> > start[2]=size[2]/2 and size[2]/=2 immediately after reading the
> image.
> > I use the Mattes MI metric and Centered Rigid 3D image
> transformation.
> > But every time I tried to run the .exe, it always told me there are
> > too many samples outside the moving image. And I choose a smaller
> > number for number of samples. I still got the same exception.
> > Why should this happen?
> > It works well before I set the region.
> > Best Regards,
> > Yu Qi
> >
> > */Luis Ibanez /* wrote:
> >
> >
> > Hi Yu,
> >
> > Yes, you can use ITK for registering a fMRI to an MRI
> > even if they dont fully cover the same anatomical
> > regions.
> >
> > You may want to use the MRI image as the Fixed image and
> > set the FixedImageRegion to the rectangular box that covers
> > the brain plus a border of about 10 millimeters. Take the
> > fMRI image as the Moving image.
> >
> >! You may want to use any of the MutualInformation Metrics
> > available in ITK.
> >
> > For a full list of the Image Metrics please look at the
> > Groups in Doxygen:
> >
> >
> http://www.itk.org/Insight/Doxygen/html/group__RegistrationMetrics.html
> >
> >
> >
> > Regards,
> >
> >
> > Luis
> >
> >
> > ------------------
> > Yu Qi wrote:
> >
> > > Hi,
> > >
> > > I have a MRI volume of the whole head and
> > > a fMRI volume of the brain.
> > >
> > > Can I register these two directly with ITK?
> > >
> > > Do I! need to use the SetRegion() to MRI volume?
> > >
> > > Thanks!
> > >
> > >
> > >
> > >
> > > *Best Regards!*
> > > **
> > > *Yu Qi*
> > >
> > >
> > >
> > >
> > >
> >
> ------------------------------------------------------------------------
> > > *D! o You Yahoo!?*
> > > 150万曲MP3疯狂搜，带您闯入音乐殿堂
> > > 美女明星应有尽有，搜遍美图、艳图和酷图
> > > 1G就是1000兆，雅虎电邮自助扩容！
> > >
> > >
> > >
> > >
> > >
> >
> ------------------------------------------------------------------------
> > >
> > > _______________________________________________
> > > Insight-users mailing list
> > > Insight-users at itk.org
> > > http://www.itk.org/mailman/listinfo/insight-users
> >
> >
> >
> >
> >
> ------------------------------------------------------------------------
> > *Do You Yahoo!?*
> > 150万曲MP3疯狂搜，带您闯入音乐殿堂
> > 美女明星应有尽有，搜遍美图、艳图和酷图
> > 1G就是1000兆，雅虎电邮自助扩容！
> >
> >
> >
> >------------------------------------------------------------------------
> >
> >_______________________________________________
> >Insight-users mailing list
> >Insight-users at itk.org
> >http://www.itk.org/mailman/listinfo/insight-users
> >
> >
>
>
> ------------------------------------------------------------------------
> *Do You Yahoo!?*
> 150万曲MP3疯狂搜，带您闯入音乐殿堂 
> 美女明星应有尽有，搜遍美图、艳图和酷图 
> 1G就是1000兆，雅虎电邮自助扩容！
> 
>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Insight-users mailing list
>Insight-users at itk.org
>http://www.itk.org/mailman/listinfo/insight-users
> 
>




---------------------------------
Do You Yahoo!?
150万曲MP3疯狂搜，带您闯入音乐殿堂
美女明星应有尽有，搜遍美图、艳图和酷图
1G就是1000兆，雅虎电邮自助扩容！
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20050426/b75bfefc/attachment.htm