That makes sense . Thanks a lot . 
Is there a way to implement this approach for my data set .
Here is an updated PR:
1 Like
That’s awesome . Can you tell me a bit more how to use this PR for my code .
Thanks a lot .
You could build from source. Or wait until this is integrated into master and makes its way into a release, so you can pip install it.
You should be able to instantiate the registration method and metric as below:
registration_metric = itk.PointSetToPointSetMetric.PSF3.New()
registration_method = itk.PointSetToPointSetRegistrationMethod.REGF3F3.New()
1 Like
Alternatively, registration_metric = itk.PointSetToPointSetMetric[PointSetType].New()
1 Like
Just for clarification . In my Bspline registration function , I should replace
registration_method = sitk.ImageRegistrationMethod()
and
registration_method.SetMetricAsMeanSquares()
with
registration_metric = itk.PointSetToPointSetMetric[PointSetType].New()
and
registration_method = itk.PointSetToPointSetRegistrationMethod.REGF3F3.New()
Thank you
Yes. But now you will need an extra call, something like registration_method.SetMetric(registration_metric).
Perfect . Thanks for the clarification . Will keep you posted with the results 
Hello
So I was working on my code with the Bspline FFD and my code was similar to the one used in the POPI model . Previously I was getting an error [ virtual sampled points must have one or more points ]
This is the part of the code:
registration_method.SetInitialTransformAsBSpline(initial_transform,
inPlace=False,
scaleFactors=[1,2,4])
registration_method.SetMetricAsMeanSquares()
registration_method.SetMetricAsJointHistogramMutualInformation()
registration_method.SetMetricSamplingStrategy(registration_method.RANDOM)
registration_method.SetMetricSamplingPercentage(0.01)
#registration_method.SetMetricFixedMask(Fixed_image_sitk_ffd_mask)
registration_method.SetShrinkFactorsPerLevel(shrinkFactors = [4,2,1])
registration_method.SetSmoothingSigmasPerLevel(smoothingSigmas=[2,1,0])
registration_method.SmoothingSigmasAreSpecifiedInPhysicalUnitsOn()
registration_method.SetInterpolator(sitk.sitkLinear)
registration_method.SetOptimizerAsLBFGS2(solutionAccuracy=1e-2, numberOfIterations=100, deltaConvergenceTolerance=0.01)
registration_method.AddCommand(sitk.sitkIterationEvent, lambda: iteration_callback(registration_method))
final_transformation = registration_method.Execute(Fixed_image_sitk_ffd, Moving_image_sitk_ffd)
print('\nOptimizer\'s stopping condition, {0}'.format(registration_method.GetOptimizerStopConditionDescription()))
so when I run this code without commenting the line registration_method.SetMetricFixedMask(Fixed_image_sitk_ffd_mask) I still get the same error as above . But when I comment this line the code runs and gives me an output of
-0.64
Optimizer’s stopping condition, Already minimized
my moving and fixed images for now are the lung dicom images that have been saved as .mha files , but I haven’t used any mask files in this code .
Can you provide some insights about what might be causing that error in the place .
The optimizer cannot improve registration metric. Try setting tighter tolerances (e.g. solutionAccuracy=1e-4).
Please repeat the error message, as this thread is already long and it is not clear to which “above error” you are referring. Also consider starting a new discussion thread.
1 Like
Sorry for not being clear in the first place .
So by error I meant
final_transformation = registration_method.Execute(Fixed_image_sitk_ffd, Moving_image_sitk_ffd)
File “C:\Users\singhgau\Anaconda3\lib\site-packages\SimpleITK\SimpleITK.py”, line 10387, in Execute
return _SimpleITK.ImageRegistrationMethod_Execute(self, fixed, moving)
RuntimeError: Exception thrown in SimpleITK ImageRegistrationMethod_Execute: d:\a\1\sitk-build\itk-prefix\include\itk-5.2\itkImageToImageMetricv4.hxx:270:
ITK ERROR: MeanSquaresImageToImageMetricv4(000001ABEF49DE90): VirtualSampledPointSet must have 1 or more points.
This is when I have registration_method.SetMetricFixedMask(Fixed_image_sitk_ffd_mask) .
1 Like
This probably means there is no overlap of your mask and your image. Potentially there is a small overlap, but none of the randomly chosen 1% of points do not fall there.
So is there a way to fix the error, or just not use this line at all .
Thanks
Try increasing sampling percentage (registration_method.SetMetricSamplingPercentage(0.01) and make sure that your fixed mask is in the same physical space as your fixed image.
1 Like
So after iterating my bspline FFD my output looks like this where I’m getting a rounding error .
14408.027
Optimizer’s stopping condition, A rounding error occurred or line-search steps have an insufficient reduction
Iteration: 50
Metric value: 14402.493636195337
My optimizer conditions are as follows
registration_method.SetOptimizerAsLBFGS2(solutionAccuracy=1e-4, numberOfIterations=100, deltaConvergenceTolerance=0.01)
Can you provide some insights of what might be causing this .
This code is equivalent to the POPI model . I tried adjusting the solution accuracy but I’m still getting this optimizer stop condition .
This might very well mean convergence has been reached.
1 Like
So I was performing the Bspline FFD registration same as the POPI model . My fixed and moving images are the two CT lung ct scans at different timepoint . And I generated 100 random fixed and moving points from the Ct images for TRE . The optimizer that I used was registration_method.SetOptimizerAsLBFGS2(solutionAccuracy=1e-2, numberOfIterations=100,deltaConvergenceTolerance=0.01) . And for metric sampling I did not used any mask as it was giving me an error of VirtualSampledPointSet must have 1 or more points.
These were the results I got which does not seem to be significant .
So my question is … Is there a way I could improve these results and is there a better way to use fixed and moving points from the images for TRE .