[Insight-users] New elastix release 4.2

[M.Staring at lumc.nl]

Dear list-subscribers,
 
We are proud to announce a new release of elastix: elastix 4.2 !
 
Elastix is a software package for image registration, based on the ITK. The software consists of a collection of algorithms that are commonly used to solve medical image registration problems. The modular design of elastix allows the user to quickly configure, test, and compare different registration methods for a specific application. In addition to the existing ITK image registration classes, elastix implements new functionality. The most important enhancements are
 
- A modular framework for sampling strategies.
- More optimisers: Kiefer-Wolfowitz, Robbins-Monro, adaptive stochastic gradient descent, evolutionary strategy.
- Complete rework of existing ITK optimisers, adding more user control and better error handling: quasi-Newton, nonlinear conjugate gradient.
- Several new or more flexible cost functions: (normalised) mutual information, implemented with Parzen windowing similar to [45], multifeature α-mutual information, bending energy penalty term, rigidity penalty term.
- The ability to concatenate any number of geometric transformations.
- The transformations support computation of not only ∂T/∂μ, but also of spatial derivatives ∂T/∂x and ∂2T/∂x2, and their derivatives to μ, frequently required for the computation of regularisation terms. Additionally, the compact support of certain transformations is integrated more generally.
- Linear combinations of cost functions, instead of just a single cost function.
 
 Elastix can be downloaded from the website, under the very open BSD license, see:
 
    http://elastix.isi.uu.nl/ <http://elastix.isi.uu.nl/> 
 
We have updated quite some components in this new release compared to elastix 4.1, and also added a few. The most important changes are:
 
- The implementation of spatial derivatives in all transform classes. This enables the computation of many regularisation (penalty) terms in a general way.
- A general way of dealing with sparse Jacobians, so that for example much less B-spline specific code is needed anymore.
- The adaptive stochastic gradient descent (ASGD) optimization routine is now usable for much larger registration problems. (It used to crash when the number of parameters was larger than 10 000.)
- We added the following components: WeightedCombinationTransform, SplineKernelTransform (which implements a ThinPlateSpline, ElasticBodySpline, VolumeSpline, and an ElasticBodyReciprocalSpline), TransformRigidityPenalty, TransformBendingEnergyPenalty, DisplacementMagnitudePenalty.
 
A more complete list of changes can be found at the website:
 
    http://elastix.isi.uu.nl/whatsnew_04_2.php <http://elastix.isi.uu.nl/whatsnew_04_2.php> 
 
If you find any bugs or have comments on this new release, please contact us through the mailinglist:
 
    elastix at bigr.nl <mailto:elastix at bigr.nl> 
 
We are also looking for a volunteer to build MacOSX binaries for us, so we can place them on the website.
 
Enjoy elastix!
 
Stefan Klein and Marius Staring 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20090818/60159ba7/attachment-0001.htm>