Building PyVE

PyVE and the packages it is made of are available as various flavours of open-source packages. Therefore if you don’t find the binary for your system (actually only a Windows binary is distributed) you can compile it yourself.

What you need is a good dose of patience.

Depending on your system however, you may or may not find suitable binary packages for the following software packages at their websites.

Prerequisites

• Python 2.x
• CMake 2.8.x (for compilation)
• SWIG 1.3.x (for the python wrapping)

Building

The following software packages must be compiled and installed on your system in order to have PyVE working.

• VTK 5.4.x
• PyVE Viewer
• ITK 3.20.x
• WrapITK 0.30.x
• GDCM (optional, reccomended to use DICOM)
• elastix required if you want to perform image registration of some kind (requires ITK 3.20)
• Qt 4.6.x (optional, reccomended for GUI development)
• SIP and PyQt4 (optional, reccomended for GUI development)
• Qwt and PyQwt (optional, reccomended for plotting in GUI)

We will cover here the compilation instructions for the first 5 packages. The other packages should compile out of the box following the general instructions that are provided at the respective websites.

The instructions are reported for Windows but the same settings can be used for the compilation in other systems. The PyVE Viewer is reported to have been compiled under Linux (Debian 6.0 powerpc) and MacOSX 10.4.11.

Building VTK

To build VTK, you need CMake.

• Download the VTK source package and extract to E:Librariesvtk-x.x.x.
• Open CMake and browse to the VTK source folder.
• Specify a location to build the binaries, for example E:Librariesvtk-x.x.x-bin.
• Click Configure, and select your target compiler, for example Visual Studio 8 2005
• Modify the following fields (some require Advanced View):

BUILD_SHARED_LIBS     ON
BUILD_TESTING         OFF
CMAKE_INSTALL_PREFIX  # Location where you want to install VTK
VTK_DATA_ROOT         # location where you downloaded and unpacked VTKData
VTK_WRAP_PYTHON       ON
VTK_USE_GUISUPPORT    ON
VTK_USE_TK            OFF

• Click ‘Configure’ again. Modify the following fields:

VTK_USE_QVTK          ON
DESIRED_QT_VERSION    4
VTK_USE_QVTK_QTOPENGL ON

• Click ‘Configure’ again.
• Click ‘Configure’ one final time, and click ‘Generate’ to create the project files.
• Open the ‘’‘Visual Studio 2005 Command Prompt’‘’ (‘’‘Start Menu|Programs|Microsoft Visual Studio 2005|Visual Studio Tools’‘’), go to the binaries directory, and build the solution:

msbuild VTK.sln /property:configuration=Release

• To install VTK to the directory specified in CMAKE_INSTALL_PREFIX, build the INSTALL project:

msbuild INSTALL.vcproj /property:configuration=Release

Qt designer plugin: To add the VTK viewer widget to the Qt designer:

• Copy [VTK]\bin\release\QVTKWidgetPlugin.dll to [Qt]\plugins\designer.

Building the PyVE Viewer

With VTK compiled and installed it is already possible to compile the PyVE Viewer class with CMake.

• Download the vtkPyveViewer source code from Sourceforge

git clone git://git.code.sf.net/p/pyve/code pyve-code

• Open CMake and browse to the vtkPyveViewer source folder.
• Specify the binaries directory,
• Click Configure
• Make sure the field CMAKE_INSTALL_PREFIX matches that of VTK.
• Click Configure, then Generate.
• Open the Visual Studio 2005 Command Prompt (Start Menu|Programs|Microsoft Visual Studio 2005|Visual Studio Tools) go to the binaries directory, and build the solution:

msbuild vtkPyve.sln /property:configuration=Release

• To install vtkPyveViewer to the directory specified in CMAKE_INSTALL_PREFIX, build the INSTALL project:

msbuild INSTALL.vcproj /property:configuration=Release

Building ITK and WrapITK

To wrap ITK for Python using the WrapITK project, you will need CableSwig. CableSwig is currently only available by CVS.

cvs -d :pserver:anonymous@public.kitware.com:/cvsroot/CableSwig co CableSwig

• Download the ITK source and extract to E:Librariesitk-x.x.x.
• Copy the CableSwig directory to [ITK]Utilities.
• Open CMake and browse to the ITK source folder.
• Specify a location to build the binaries, for example E:Librariesitk-x.x.x-bin.
• Click Configure, and select your target compiler, for example Visual Studio 8 2005
• Modify the following fields (some require Advanced View

BUILD_EXAMPLES        OFF
BUILD_SHARED_LIBS     ON
BUILD_TESTING         OFF
CMAKE_INSTALL_PREFIX  # Location where you eventually want to install ITK
ITK_USE_REVIEW        ON
USE_WRAP_ITK          ON

• Add /bigobj to the field CMAKE_CXX_FLAGS.
• Click ‘Configure’ again. Modify the following fields:

WRAP_ITK_DIMS            2;3;4
WRAP_ITK_PYTHON          ON
WRAP_Iterators           ON
WRAP_rgb_unsigned_char   ON
WRAP_rgb_unsigned_short  OFF
WRAP_rgba_unsigned_char  ON
WRAP_rgba_unsigned_short OFF
WRAP_signed_short        ON
WRAP_unsigned_char       ON
WRAP_unsigned_short      OFF

• Click Configure one final time, and click Generate to create the project files.
• Open the Visual Studio 2005 Command Prompt (Start Menu|Programs|Microsoft Visual Studio 2005|Visual Studio Tools), go to the binaries directory, and build the ITK solution:

msbuild ITK.sln /property:configuration=Release

• To install ITK to the directory specified in CMAKE_INSTALL_PREFIX, build the INSTALL project:

msbuild INSTALL.vcproj /property:configuration=Release

Building GDCM

To build GDCM, you need SWIG.

• Download the GDCM package and unpack it for example in E:\Libraries\gdcm-x.x.x
• Open CMake and browse to E:\Libraries\gdcm-x.x.x.
• Specify the binaries directory, preferably E:\Libraries\gdcm-x.x.x-bin.
• Click Configure.
• Modify the following fields:

CMAKE_INSTALL_PREFIX      # Location where you want to install GDCM
GDCM_BUILD_SHARED_LIBS    ON
GDCM_WRAP_PYTHON          ON
SWIG_EXECUTABLE           [SWIG]\swig.exe

• Click Configure, then Generate.
• Go to the binaries directory and open the solution file (GDCM.sln), and build the solution.
• To install GDCM to the directory specified in CMAKE_INSTALL_PREFIX, build the INSTALL project.

See Configuring and Building GDCM

External Binary and Python Packages

We suggest to use the packages:

• numpy/scipy
• pyDicom 0.9.5