The Insight Software Consortium is happy to announce the release of ITK 4.3.0!
The 4.3.0 release is a major milestone that marks the hard work on many outstanding community members.
We are excited to announce that experimental DICOM image reading is available via the DCMTK library as a backend in addition to the GDCM library. By default, DCMTK ImageIO support is not enabled. To try DCMTK ImageIO support, turn the CMake option Module_ITKIODCMTK ON. For Unix platforms, the supporting DCMTK library will automatically be built as a CMake ExternalProject. On Windows, the DCMTK library must be built independently of the ITK build system. Then, specify the location to the external build after setting the CMake option ITK_USE_SYSTEM_DCMTK ON.
WrapITK has undergone extensive improvements and bug fixes. The status of WrapITK on various platforms can be found on the following WrapITK Status page.
A number of registration-related feature enhancements were added. The image registration methods have been updated to accommodate multiple image metrics within a single optimization scheme. This permits, for example, registration of a T1/T2 fixed image pair with a T1/T2 moving image pair using a single metric for both T1 and T2 gradients or even using two different metrics. New exponential transforms are available along with automated B-Spline transform scale estimation. A new physics-based non-rigid registration class is also available.
Updated third-party library versions include:
- NrrdIO from Teem 1.11,
- libtiff 4.03,
- FFTW 3.3.2
A large number of important bugs were crushed, including fixes to the ObjectFactoryBase, Patch-Based Denoising improvements, support for a system libtiff, mesh processing, and FFTW use.
Support for itk::VectorImage’s was added to many algorithms, and a single templated Adaptor class can now be applied without modification to both itk::VectorImage’s and itk::Image’s of itk::Vector’s, itk::RGBPixel’s, etc.
An extensive amount of code cruft was removed and coding style was made more consistent; ITK gets more stylish and classy with every release!
Performance optimizations were achieved by removing GetInput/GetOutput calls within filter inner loops, improving memory alignment, memory access pattern optimizations, and parallelization through threading.
NOTE: As previously scheduled, Visual Studio 2005 will no longer be supported after this release.