Multi-lingual medical image registration library
C++ CMake TeX Python Shell R Other
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.
.ExternalData Create separate source and data tar-balls for distribution. Jun 10, 2015
CMake Merge tag 'v1.0.1' into develop Aug 5, 2017
Code ENH: Remove ExecuteInverse Oct 31, 2017
Documentation DOC: Add guide for installing SimpleElastix after compilation has fin… Aug 4, 2017
Examples Fixes on the sample java code due to changes of the newly released ve… Oct 30, 2017
ExpandTemplateGenerator Fix spelling errors Jun 6, 2017
SuperBuild ENH: Update elastix Nov 8, 2017
Testing ENH: Remove ExecuteInverse Oct 31, 2017
Utilities Add variable options to manylinux docker builds Apr 6, 2017
Wrapping ENH: Import elastix the same way ITK is imported Oct 18, 2017
.gitattributes Modify git attributes to allow big swig dot i files Dec 17, 2012
.gitignore Merge branch 'rc3' into develop Apr 3, 2017
.hooks-config.bash ENH: Simplify local Git hook chaining Feb 13, 2012
.travis.yml ENH: Update Elastix Oct 30, 2017
CMakeLists.txt ENH: Import elastix the same way ITK is imported Oct 18, 2017
CTestConfig.cmake Updating Dash info Jun 21, 2014
LICENSE ENH: Adding the Apache license. Nov 25, 2010
NOTICE ENH: Update copyright notice Mar 10, 2016 DOC: Remove incomplete installation instructions from and o… May 3, 2017 ENH: Import elastix the same way ITK is imported Oct 18, 2017 Require SimpleITK major version to match for compatible version Apr 8, 2014 Remove required flags and replace with compile options property Feb 22, 2017
Version.cmake Bumping version 1.0.1 for patch release Aug 2, 2017
circle.yml Make the release branch also continuous May 15, 2017

What is SimpleElastix?

Image registration is the process of transforming images into a common coordinate system so corresponding pixels represent homologous biological points. SimpleElastix is an extension of SimpleITK that offers a user-friendly API to the popular image registration algorithms of the elastix C++ library. This makes state-of-the-art medical image registration really easy to do in languages like Python, Java, C# and R. This package provides

  • elastix and transformix bindings for C++, Python, Java, R, Ruby, Octave, Lua, Tcl and C# (see elastix manual for a list of supported registration algorithms).
  • Pre-configured registration methods that work well in many cases and serve as starting points for tuning elastix to domain-specific applications.
  • Installation guides, examples, and introductory material at
  • A user-friendly API that aligns with the design philosophy of SimpleITK developed specifically for rapid prototyping. If you are interested, The Design of SimpleITK is a great read.
  • The complete set of SimpleITK image processing algorithms.

Enough talk, time for some examples! We will use Python for the following code. Say you need to register two images. This can be accomplished with a single line of code:

resultImage = SimpleITK.Elastix(sitk.ReadImage("fixedImage.dcm"), sitk.ReadImage("movingImage.dcm"))

Under the hood, Elastix will use stochastic optimization for maximum speed, a multi-resolution strategy and several different transforms of increasing complexity for maximum robustness. All aspects of the registration procedure can be customized via parameter maps.

SimpleElastix can also be used for more complex image processing pipelines. Say you want to compare the volume, mean intensity and standard deviation of the intensity of anatomical structures across a population of images using an atlas segmentation. We can accomplish this task with the following lines of Python code:

import SimpleITK as sitk

# The atlas and associated segmentation is loaded once and held in memory
movingImage = sitk.ReadImage('atlasImage.hdr')
movingLabel = sitk.ReadImage('atlasLabel.hdr')

# Images are loaded from disk one at a time. Here we specify an array of 
# paths to images which we will loop over. 
population = ['image1.dcm', 'image2.dcm', ... , 'imageN.dcm']

selx = sitk.SimpleElastix()

for filename in population
  # Register images
  fixedImage = sitk.ReadImage(filename)

  # Transform label map using the deformation field from above
  resultLabel = sitk.Transformix(movingLabel, selx.GetTransformParameterMap())

  # Compute statistics for label 1
  LabelStatistics = sitk.LabelStatisticsImageFilter()
  LabelStatistics.Execute(fixedImage, sitk.Cast(resultLabel, sitk.sitkInt8))
  # etc etc

This example demonstrates the efficiency of combining SimpleElastix's object oriented interface (the way we used elastix to register images) and procedural interface (the way we used transformix to warp labels) with SimpleITK (the way we computed statistics). Previously, using elastix and transformix on large datasets would incur a significant overhead, from scripting command line invocations and arguments to copying images and transform parameter files across folders. With SimpleElastix this complexity is easier to manage and more memory and disk I/O efficient. For more examples see the documentation or the Examples/SimpleElastix directory.

Building with the SuperBuild

SimpleElastix integrates elastix and transformix with the SimpleITK SuperBuild. Simply clone this repository and invoke the SuperBuild as outlined in the documentation. The SuperBuild will download and install dependencies (elastix, ITK, SimpleITK and SWIG) and compile SimpleElastix. Target language dependencies need to be pre-installed, e.g. sudo apt-get install cmake swig monodevelop r-base r-base-dev ruby ruby-dev python python-dev tcl tcl-dev tk tk-dev. Note that this project takes around an hour to build on a quad-core machine.

The documentation further describes how to build SimpleElastix on Windows and how to build SimpleElastix manually without the SuperBuild.

SimpleElastix has been tried and tested on Ubuntu 14.10 using GCC 4.9.2 and Clang 3.4.0, Mac OSX Yosemite using Apple Clang 600.0.56 and Windows 8.1 using Microsft Visual Studio 2012 C++ compiler.


If you are interested in my work you are most welcome to visit my website.