[ITK-users] How to use a binary mask in a registration process ?

[neon29]

Bradley Lowekamp wrote
> Hello,
> 
> First verify that you are using the correct methods for setting the masks.
> You wrote "SetMetricMovingImage" where I think you intended
> "SetMetricMovingMask". Mistakes are easy to make in code and things like
> this should be double/triple checked.
> 
> You likely need to perform a morphological erosion on the masks. With
> smoothing and derivative computation, values outside the mask can effect
> the metric near the boarder. You need to estimate how big the radius of
> the erosion need to be based on your smoothing sigmas, and your scale.
> 
> Without sample code I don't think I can help further.
> 
> HTH,
> Brad
> 
> 
> On Sep 14, 2015, at 11:11 AM, neon29 <

> florian.nicolas@

> > wrote:
> 
>> Hi everybody!
>> 
>> I am currently trying to register sonar images using SimpleITK with
>> Python.
>> As the two images I want to register are geo-projected, I would like to
>> restrict each image to the pixels where I have data (and get rid of the
>> "background" which consists in '0').
>> 
>> I have already computed my two binary masks and I have tried to use it
>> with
>> the SetMetricFixedMask() and SetMetricMovingImage() methods but when I
>> look
>> at the result, it is the same as when I do not use any mask. Indeed, it
>> seems that the registration process tend to overlap my masks.
>> 
>> I hope someone could help!
>> 
>> Florian
>> 
>> 
>> 
>> --
>> View this message in context:
>> http://itk-insight-users.2283740.n2.nabble.com/How-to-use-a-binary-mask-in-a-registration-process-tp7587799.html
>> Sent from the ITK Insight Users mailing list archive at Nabble.com.
>> _____________________________________
>> 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://www.kitware.com/products/protraining.php
>> 
>> 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://public.kitware.com/mailman/listinfo/insight-users
> 
> _____________________________________
> 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://www.kitware.com/products/protraining.php
> 
> 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://public.kitware.com/mailman/listinfo/insight-users

Thanks for your reply!

Here is the piece of code:

fixed = sitk.ReadImage('#######/imgRef.png', sitk.sitkFloat32);
moving = sitk.ReadImage('#######/imgRep.png', sitk.sitkFloat32)
maskFixed = sitk.ReadImage('#######/maskRef.png', sitk.sitkUInt8)
maskMoving= sitk.ReadImage('#######/maskRep.png', sitk.sitkUInt8)

# Handle optimizer
R = sitk.ImageRegistrationMethod()

# Restrict the evaluation of the similarity metric thanks to masks
R.SetMetricFixedMask(maskFixed)
R.SetMetricMovingMask(maskMoving)

# Set metric as mutual information using joint histogram
R.SetMetricAsMattesMutualInformation(numberOfHistogramBins=255)

# Gradient descent optimizer
R.SetOptimizerAsRegularStepGradientDescent(learningRate=0.01, minStep=1e-5,
numberOfIterations=100, gradientMagnitudeTolerance=1e-8)

#R.SetOptimizerScalesFromPhysicalShift()

R.SetMetricSamplingStrategy(R.REGULAR) #R.RANDOM

# Define the transformation (Rigid body here)

transfo = sitk.CenteredTransformInitializer(fixed, moving,
sitk.Euler2DTransform())

R.SetInitialTransform(transfo)

# Define interpolation method
R.SetInterpolator(sitk.sitkLinear)

# Add command to the registration process
R.AddCommand(sitk.sitkIterationEvent, lambda: command_iteration(R))
R.AddCommand(sitk.sitkStartEvent, lambda: start_plot())
R.AddCommand(sitk.sitkEndEvent, lambda: end_plot())
R.AddCommand(sitk.sitkIterationEvent, lambda: current_plot(R))
# Perform registration
outTx = R.Execute(fixed, moving)

print(outTx)
print("--------")
print("Optimizer stop condition:
{0}".format(R.GetOptimizerStopConditionDescription()))
print("Number of iterations: {0}".format(R.GetOptimizerIteration()))
print("--------")

# Perform transformation and resample the moving image

# Save transformation as tfm file
sitk.WriteTransform(outTx,
'/home/egs/f_nicolas/CODES_THESE/transfo_final.tfm')
#sitk.Show(transfo.GetDisplacementField(),"Displacement field")

# Resample moving image according to the last transformation
resampler = sitk.ResampleImageFilter()
resampler.SetReferenceImage(fixed)
resampler.SetInterpolator(sitk.sitkLinear)
#resampler.SetDefaultPixelValue(100)
resampler.SetTransform(outTx)
out = resampler.Execute(moving)






--
View this message in context: http://itk-insight-users.2283740.n2.nabble.com/How-to-use-a-binary-mask-in-a-registration-process-tp7587799p7587809.html
Sent from the ITK Insight Users mailing list archive at Nabble.com.