[ITK-users] What is an accepted duration of a 3D registration? Should I expect the MSE to "converge" to 0?

Marina Bendersky marina.bendersky at gmail.com
Mon Sep 14 18:56:08 EDT 2015


Hello,

I'm doing a 3D registration of 2 CT volumes (the fixed one contains
131 dicom images and the moving has 145 dicom images).

I first converted each dicom series into one mha volume, using the
example DicomSeriesReadImageWrite2.cxx and then, I do the registration
following the example imageRegistration20.cxx. Below are the main
registration components and parameters (other parameters had their
default values):

typedef itk::AffineTransform< double, Dimension  >  TransformType;
typedef itk::RegularStepGradientDescentOptimizer     OptimizerType;
typedef itk::MeanSquaresImageToImageMetric< FixedImageType,
MovingImageType >  MetricType;
typedef itk::LinearInterpolateImageFunction< MovingImageType, double >
 InterpolatorType;

double translationScale = 1.0 / 1000.0;
optimizer->SetMaximumStepLength( 0.05 );
optimizer->SetMinimumStepLength( 0.01 );


This 3D registration runs on a MacBookPro laptop for 5.5 hours, so I
was wondering, is this "normal"? How long should a 3D registration
take?

Also, the MSE metric keeps decreasing (though incrementally) even
after 300 iterations (plot attached), and there are oscillations
around the iteration 150 (I ran this twice and both times I get this
plot!) - why could this be?

Should I expect the MSE metric to converge to 0, or something closer
to 0 (instead of the current values, in the ninety thousands...)?

Thank you for your help,
Marina
-------------- next part --------------
A non-text attachment was scrubbed...
Name: metricMSE300Iters.png
Type: image/png
Size: 13390 bytes
Desc: not available
URL: <http://public.kitware.com/pipermail/insight-users/attachments/20150914/e4b973db/attachment-0001.png>


More information about the Insight-users mailing list