Skip to content

Edit this page

Open and issue

Installing NiftyReg


I like to build within a directory named code found in the root of my user directory, to create this directory run:

mkdir ~/code

Now create a build, install and source directory for NiftyReg:

cd ~/code
mkdir niftyreg_{build,source}

Obtain the newest version of NiftyReg or by running the following:

cd ~/code/niftyreg_source
git clone .

Run CMake/Make


If you don't have make/cmake, follow the steps in this guide.

Make the build and install directories.

The default install location is /usr/local, which falls on your PATH. This is the easiest location to install. If you want to install somewhere else then specify the path in the CMAKE_INSTALL_PREFIX variable.

cd ~/code/niftyreg_build
ccmake ~/code/niftyreg_source

If you get an error like: Error opening terminal: xterm-256color. run the following and try again (you can also add this to your .bashrc file):

export TERM=xterm

The following options will be displayed, ensure you change CMAKE_INSTALL_PREFIX variable to the install directory path if you don't want to use default:

Parameter Value
BUILD_ALL_DEP Set to ON if you want to build All the dependencies
BUILD_SHARED_LIBS Build the libraries as shared build the libraries as shared
BUILD_TESTING Set to ON if you want to build the unit tests
CMAKE_BUILD_TYPE Compiling options: Debug Release RelWithDebInfo MinSizeRel
CMAKE_INSTALL_PREFIX Set the path where the final install will be copied
M_LIBRARY Path to a library.
PNG_INCLUDE_DIR Set if you want NiftyReg to support the PNG file format for 2D images. Note that CMake will try to find the libpng on your system and will build it automatically if it does not find it.
USE_CUDA Set to ON if you want to build the GPU code. The CUDA toolkit must be install otherwise CMake will return an error message
USE_OPENCL Set to ON to use OpenCL for multi-CPU implementation.
USE_OPENMP Set to ON to use OpenMP for multi-CPU implementation.
USE_SSE Set to ON to use SIMD based implementation, mostly for cubic B-Spline related computation. Note that SIMD implementation has only be done for single precision.

Press 'c' to configure the NiftyReg project, press 'c' to configure the project. Once the project is correctly configured, press the 'g' key to generate the Makefiles. You can then build and install the project:

sudo make -j4
sudo make install

If you get an error try setting USE_OPENMP to OFF while running ccmake.

Post-Install Configuration

The project should then be installed into the CMAKE_INSTALL_PREFIX directory you previously created.

If you changed the install prefix directory, to one not on your PATH, you will need to add NiftyReg to your PATH. In order to use NiftyReg in any terminal, you will need to edit your .profile file by adding the following lines:

export PATH={NREG}/bin:${PATH}

Close and re-open the linux terminal then run: