[Elastix] problems with affine

Esmeralda Ruiz Pujadas esmeralda.ruizpujadas at gmail.com
Tue Feb 3 17:54:05 CET 2015


Dear Floris,

You are right, the problem is that it makes a global scale and I am working
in 3d so
I cannot increase the size separately. I also get shifted in z and in the
first frame
appears nothing even when there was a very similar shape. I wonder if
the best option is to go slice by slice.

Thank you so much

The parameter file
//ImageTypes
(FixedInternalImagePixelType "float")
(FixedImageDimension 3)
(MovingInternalImagePixelType "float")
(MovingImageDimension 3)

//Components
(Registration "MultiMetricMultiResolutionRegistration")
(FixedImagePyramid "FixedSmoothingImagePyramid")
(MovingImagePyramid "MovingSmoothingImagePyramid")
(Interpolator "BSplineInterpolator")
(Metric "AdvancedNormalizedCorrelation" "TransformBendingEnergyPenalty")
(Optimizer "StandardGradientDescent")
(ResampleInterpolator "FinalBSplineInterpolator")
(Resampler "DefaultResampler")
(Transform "AffineTransform")
// ********** Pyramid

// Total number of resolutions
(NumberOfResolutions 4)


// ********** Transform

//(Scaler 500000.0)
//(CenterOfRotation 128 128 ??) center by default
(AutomaticTransformInitialization "true")
(HowToCombineTransforms "Compose")


// ********** Optimizer

// Maximum number of iterations in each resolution level:
(MaximumNumberOfIterations 600 600 600 900)

//SP: Param_a in each resolution level. a_k = a/(A+k+1)^alpha
(SP_a 1000.0 1000.0 500.0 500.0)

//SP: Param_alpha in each resolution level. a_k = a/(A+k+1)^alpha
(SP_alpha 0.602)

//SP: Param_A in each resolution level. a_k = a/(A+k+1)^alpha
(SP_A 50.0)


// ********** Metric

//Number of grey level bins in each resolution level:
(NumberOfHistogramBins 32)
(FixedLimitRangeRatio 0.0)
(MovingLimitRangeRatio 0.0)
(FixedKernelBSplineOrder 1)
(MovingKernelBSplineOrder 3)


// ********** Several

(WriteTransformParametersEachIteration "false")
(WriteTransformParametersEachResolution "false")
(ShowExactMetricValue "false")
(ErodeMask "true")
(UseDifferentiableOverlap "false")


// ********** ImageSampler

//Number of spatial samples used to compute the mutual information in each
resolution level:
(ImageSampler "RandomCoordinate")
(NumberOfSpatialSamples 3000)
(NewSamplesEveryIteration "true")



// ********** Interpolator and Resampler

//Order of B-Spline interpolation used in each resolution level:
(BSplineInterpolationOrder 1)

//Order of B-Spline interpolation used for applying the final deformation:
(FinalBSplineInterpolationOrder 0)

//Default pixel value for pixels that come from outside the picture:
(DefaultPixelValue 0)


// Choose whether to generate the deformed moving image.
// You can save some time by setting this to false, if you are
// not interested in the final deformed moving image, but only
// want to analyze the deformation field for example.
(WriteResultImage "true")


On Tue, Feb 3, 2015 at 5:02 PM, Floris Berendsen <floris at isi.uu.nl> wrote:

> Hi Esmeralda,
>
> The large object will indeed affect the affine registration, since affine
> is a global transform (i.e. its scaling, shearing, etc. applies in the same
> amount everywhere in the image).
> I do expect a better registration too when you removed the large object
> (you made the intensities similar to the background?). Can you provide more
> information about your registration settings? I.e. what metric do you use,
> what is the pyramid schedule, etc. The easiest is to send your parameter
> file. And what are the sizes of you images (voxels and/or millimeters). Or
> can it be that the small circle is only a cross section of a bigger
> structure in 3d, which are registered to each other?
>
> Best,
> Floris
>
> On Tue, Feb 3, 2015 at 1:34 PM, Esmeralda Ruiz Pujadas <
> esmeralda.ruizpujadas at gmail.com> wrote:
>
>> Dear all,
>>
>> I have a doubt as shown in the image. I am using an affine registration
>> and
>> I have a 3d volume as shown in the image with fixed, moving and result,
>> respectively.
>> I would expect that the result of the small circle (third) to grow and be
>> scaled
>> as the small circle in the first image as it is larger. And the
>> registration does not
>> do any transformation instead. Why is it not scaling?. Is it affected by
>> the big circle?.
>> I also try to obtain the small circle and delete the big one but even
>> like this, I
>> did not find any result of the affine registration.
>>
>> Thanks
>> Sincerely
>> Esmeralda
>>
>> _______________________________________________
>> Elastix mailing list
>> Elastix at bigr.nl
>> http://lists.bigr.nl/cgi-bin/mailman/listinfo/elastix
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bigr.nl/pipermail/elastix/attachments/20150203/d0e08003/attachment.html>


More information about the Elastix mailing list