[Insight-users] Problem with output of rigid registration of different sized 3d volumes

[Karthik Krishnan]

We've registered PET-CT scans of rats, since the images have different 
resolutions and the PET image in physical space is much larger than the 
CT data.

But once you get the transform resulting from the registrtation, its 
entirely up to you how you use the transform. For instance we used the 
inverse transformation to map the fixed image onto the moving image (for 
overlays). since you are using rigid registration, the inverse is well 
defined.

Setting the extents of the resample filter should do the filling for you.

-karthik

sk012s0481_2 at blueyonder.co.uk wrote:

>We're working on 3d rigid registration using the example ITK code. We want
>to register two datasets of different dimensions and obtain an output 3D
>volume that encompasses both datasets. Our code is importing 2D DICOM
>slices, and then exporting them as either one 3D dataset or another 2D
>DICOM dataset.
>
>We're currently finding the following problems:
>
>
>1) If we export our registered result as one dataset, the dimensions of
>the final result match the dimensions of our fixed image which we're
>registering to. If the moving volume is smaller than the fixed volume then
>this leads to loss of information, which we don't want.
>
>e.g. When registering a 200x100x100 dataset to a 150x100x100 dataset, the
>final result is 150x100x100, where we lose information from the other
>dataset.
>
>If we perform the registration the other way round, ITK produces a
>200x100x100 dataset which fills in the non-existing slices in the moving
>image with zero values. This is the behaviour we want when the dimensions
>of the smaller dataset are smaller.
>
>
>
>2) When doing a registration of a 150x100x50 dataset onto a 150x100x100
>dataset, if we try and export as another DICOM series, the code crashes,
>saying the following:
>
>===========
>Exception thrown while writing the series
>itk::ExceptionObject (0126F0A8)
>Location: "Unknown"
>File: ...\insighttoolkit-2.4.0\code\io\itkImageSeriesWriter.txx
>Line 312
>Description: itk::ERROR: ImageSeriesWriter(018D1FE8): The number of
>filenames passed is 50 but
>
>100 were expected.
>=========
>
>It doesn't generate a result with the non-existant slices being filled in,
>as is the case with the single file output code. Are there any ways we can
>write out a DICOM series which does fill in the missing slices, and which
>ultimately samples the overall volume of the entire registration result?
>
>Thanks in advance for your time,
>
>Rehan Ali
>
>
>
>
>_______________________________________________
>Insight-users mailing list
>Insight-users at itk.org
>http://www.itk.org/mailman/listinfo/insight-users
>
>  
>