Template Image Processing Library (TIPL)

Introduction

Template Image Processing Library (TIPL) is a lightweight C++ template library designed mainly for high-performance medical imaging processing.

To use it, include the root header tipl.hpp

#include "TIPL/tipl.hpp"  

The library supports Linux, MacOS, Windows, and Jupyter Notebook

Installation with CMake

While the library can be used as is, providing the install location is specified to using projects, we provide a CMake system based installation. To install with CMake do the following:

• Download the source to a directory called ./TIPL
• Configure the installation:

  cd TIPL; mkdir build; cd build
  cmake .. 
  cmake --build . 
  cmake --install . --prefix <install_location>

  or alternatively if using the make build system:

  cd TIPL; mkdir build ; cd build
  cmake .. -DCMAKE_INSTALL_PREFIX=<install_location>
  make 
  make install

Using the installed packge

Afterwards client software building with CMake can use this package by specifying

find_package(TIPL)

in its CMakeLists.txt and linking the client application to the imported target TIPL::tipl

A fly in the ointment is that I do not currently know how to make the TIPL libraries C++ standard (14) to be propagated. Client applications should also set the CXX_STANDARD property of downstream libraries and executables to 14.

When the client is configured with CMake, it is necessary for CMake to be able to find the installed TIPL this can be done by adding <install_location> for TIPL to the CMAKE_PREFIX_PATH or by explicitly specifying TIPL_DIR as

cmake -DTIPL_DIR=<install_dir>/lib/cmake/TIPL

Example

• Notebooks examples:

  • Image IO
  • Volume and Slicer Operations
  • Pixel Operations (Filters)
  • Morphological Operations
  • NIFTI file viewer

• Google colab examples:

  • Load NIFTI file
  • Image registration

Design paradigm

The design paradigm is to provide an "easy-to-use" and also "ready-to-use" header-only library. Most existing image processing libraries are not designed for medical image analysis and have suboptimal performance. TIPL was developed to fill this gap. The design paradigm follows several coding guidelines and principles that make it highly efficient and reusable.

The following is the main paradigm behind TIPL:

• Decouple image type and image processing method: Most of the image processing libraries are limited to their defined image type. TIPL is not. You may use a pointer or any kind of memory block to as the input. This reduces unnecessary memory storage and copy.

• Not limited to RGB pixel type: In medical imaging, the most common pixel type is "short" or "float", not the RGB value. TIPL makes no assumption on the pixel type to achieve the best applicability.

• Minimize class inheritance: Class inheritance is known to cause difficulties in code maintenance and unfriendly for extensions. TIPL uses template variable coupling to minimize the need for inheritance unless a strong case for inheritance is suggested. This provide a "flat" library structures that is easy to maintain and modify.

• Minimize between-header dependency: TIPL couples function and type at very late stage in the cpp to reduce header dependency. This allows for fast compilation time and achieve better abstraction.

• Avoid using plateform dependency functions or syntax: TIPL will be compiled against MSVC, CLANG, and GCC to ensure the best cross-platform capability.

Installation

TIPL is header only. You may clon the repo or use CMake to install the package.

1. Get header files from Github

!git clone http://github.com/frankyeh/TIPL/

2. Include the header

#include "TIPL/tipl.hpp"  

Now you can use TIPL

Building the C++ examples using CMake

It is possible to install TIPL using CMake, in which case C++ packages using CMake can easily build with it. In this package C++ examples are available in the cpp subdirectory. To build these using CMake:

• Make sure TIPL is installed with CMake, e.g. in a directory called <TIPL_install_directory> (see the TIPL README.md file how to do this).
• Execute the commands from the root of your cloned TIPL-example directory:

  export CMAKE_PREFIX_PATH=<TIPL_install_directory>:$CMAKE_PREFIX_PATH
  mkdir build ; cd build
  cmake ..
  cmake --build .
  cd cpp

• At this point one should be able to execute the exmaple, e.g.:

  ./linear_reg

• We also made a directory cpp/TIPL-examples and symbolic linked the source data directory there so that the exmaples can find their inputs.

Naturally one can also locate the CMake TIPLConfig.cmake file using other CMake means than CMAKE_PREFIX_PATH e.g. by providing the CMake option: -DTIPL_DIR=<TIPL_install_directory>/lib/cmake/TIPL to cmake. One should also be able to build out of source (the build directory can really be anywhere).

Installation of the examples is not yet supported