On behalf of the Insight Segmenation and Registration Toolkit community, we present the first release candidate for ITK 4.10. The release features a new remote module, TwoProjectionRegistration, which can simultaneously register two projection images to a three-dimensional image volume. TwoProjectionRegistration has applications in external beam radiotherapy among other areas of medical imaging. The module uses Powell’s method for optimization and the Siddon-Jacobs algorithm for rapid ray tracing.
ITK 4.10 also updates the internal numerics library, VNL. ITK now pushes patches upstream and syncs through a Git subtree workflow. In other updates, Windows builds processes >4GB images by default, and an updated GDCM has better Digital Imaging and Communications in Medicine (DICOM) support.
In regards to Python wrapping, ITK 4.10 removes the requirement to specify type during the creation of most filters. Alternatively, ITK infers the required type based on a primary input that it obtains during filter creation. Take the below explicit type specification as an example:
median = itk.MedianImageFilter[ImageType, ImageType].New()
It is now possible to call the following:
median = itk.MedianImageFilter.New(Input=reader.GetOutput())
It is also possible to call shortened versions:
- median = itk.MedianImageFilter.New(reader.GetOutput())
- median = itk.MedianImageFliter.New(reader)
git clone http://itk.org/ITK.git
git checkout -q –detach v4.10rc01
For more details, please see the ITK Git documentation.
We appreciate the efforts of the 26 contributors to this release. We would especially like to recognize new contributors: Andrey Fedorov, Sujin Philip, Francois Budin, and Alvaro Sanchez. The 4.10.0 final release is scheduled for the end of May. Stay tuned for updates on future release candidates!