Insight Toolkit (ITK)

The gdcmjpeg directory contains a specialized subset of the Independent JPEG Group's (IJG) libjpeg version 6b. This subset is tailored for DICOM implementations and includes modifications to support various bit depths and lossless image reading.

Key characteristics:

• Build System: The standard libjpeg build process is replaced with a CMake build process.
• Exclusions: Standard executables like cjpeg and djpeg are not included.
• Bit Depth Support: The distribution is modified to allow the generation of 8-bit, 12-bit, or 16-bit JPEG libraries.
• DICOM Compatibility: Includes patches (such as the 'well known' 12 bits Philips MRI DICOM patch) to handle specific JPEG compressions found in DICOM.

ITK includes a subset of the OpenJPEG project, which is an open-source JPEG 2000 codec written in C. This library is used to support the JPEG 2000 still-image compression standard.

Note that ITK only includes the necessary components to build libopenjpeg. It does not include the full OpenJPEG distribution (such as the codec subdirectories) and uses a CMake-based build process instead of the standard OpenJPEG build process.

ITKproxTV is a third-party ITK module that provides proxTV, a toolbox of fast total variation proximity operators. It is specifically designed to support the TotalVariation module within ITK.

Key characteristics:

• Automatic Build: The library source is not part of the ITK tree. It is automatically cloned and built from the InsightSoftwareConsortium proxTV fork during the ITK build process using ExternalProject.
• On-demand Configuration: The module is marked as EXCLUDE_FROM_DEFAULT. It is only configured and built if a dependent module (such as TotalVariation) is enabled in your ITK configuration.
• Version Pinning: The specific version used is managed via proxTVGitTag.cmake.

The TextureFeatures module provides N-dimensional textural feature image filters. These filters compute texture features over a sliding window for each voxel of an input image, generating per-feature output maps. These maps are designed for use in radiomics and computer-vision pipelines.

Key capabilities include computing:

• First-order features
• Run-length features
• Co-occurrence (GLCM) features

The Insight Toolkit (ITK) is an open-source, cross-platform toolkit designed for N-dimensional scientific image processing, segmentation, and registration.

Key Capabilities

• Segmentation: Identifying and classifying data within a digitally sampled representation (e.g., images from CT or MRI scanners).
• Registration: Aligning data or developing correspondences between different datasets (e.g., aligning a CT scan with an MRI scan to combine information).

ITK uses an open governance model and is fiscally sponsored by NumFOCUS.

The ITK repository includes a specialized fork of the Eigen library. This fork is modified to allow Eigen to be embedded directly into the ITK source tree. The primary modifications include:

• Build System Integration: The CMake build system has been modified to integrate Eigen with ITK's module system.
• Symbol Mangling: A header is provided to facilitate the mangling of symbols. This ensures that Eigen's symbols live within an itkeigen namespace, preventing symbol conflicts if other copies of the Eigen library are loaded within the same process.

CharLS is an optimized implementation of the JPEG-LS standard (ISO-14495-1/ITU-T.87) designed for lossless and near-lossless image compression. It is characterized by low complexity (fast compression) and high compression ratios comparable to JPEG 2000. CharLS is intended to outperform other open-source and commercial JPEG-LS implementations in terms of speed.

Note: The version included in this directory is a subset of the CharLS 2.0.0 release, specifically containing only the src/ subdirectory and the original CMakeLists.txt, with Windows-specific compilation files removed.

The PrincipalComponentsAnalysis module provides ITK filters for performing principal component analysis on several data types:

• Scalar data
• Vector data
• Mesh-vertex data

The flagship implementation is itk::VectorFieldPCA, which handles ensembles of vector fields on a common mesh.

The Montage module provides tools for mosaic-stitching and 3D reconstruction of large datasets from collections of partially-overlapping 2D slices. It is built around phase correlation image registration.

Core classes include:

• itk::PhaseCorrelationImageRegistrationMethod
• itk::PhaseCorrelationOperator
• itk::PhaseCorrelationOptimizer
• itk::TileMontage
• itk::TileMergeImageFilter
• itk::TileConfiguration
• itk::NMinimaMaximaImageCalculator

The pygccxml module in ITK is a reduced distribution of the pygccxml source tree containing only the library source. It is a pure Python module and does not require compilation.

Key usage details:

• Purpose: It is used by WrapITK to build C++ wrappings.
• Distribution: ITK ships its own version of pygccxml to avoid requiring users to install it as a system dependency.
• Constraint: ITK does not currently allow the use of a system-installed pygccxml; it relies exclusively on the version provided within this module.