DOCSEARCHbyAlgolia
Submit Docs

ITK-Wasm (Insight Toolkit WebAssembly)

repository·main·Indexed 42 days ago

https://github.com/insightsoftwareconsortium/itk-wasm

Enables high-performance spatial analysis by bringing the Insight Toolkit (ITK) and C++ code to WebAssembly. It supports execution in web browsers, Node.js, Python, and WASI runtimes, providing efficient data bridging and parallel pipeline execution. Includes tools like itk-convert for image and mesh format transcoding, as well as packages for regression testing such as @itk-wasm/compare-images and @itk-wasm/compare-meshes.

Tokens
212.3K
Snippets
664
Records
943
Agent score
53%

What's inside ITK-Wasm

  1. Overview of creating WebAssembly scientific C++ processing pipelines

    main
    This tutorial series covers the end-to-end workflow for developing scientific C++ processing pipelines compiled to WebAssembly (Wasm). The process involves setting up a build system, creating C++ pipelines, and executing them across different environments including the command line, Node.js, and the browser. It also covers advanced topics like handling complex data types and debugging Wasm pipelines.

  2. Overview of ITK-Wasm

    main

    ITK-Wasm combines the Insight Toolkit (ITK) with WebAssembly to enable high-performance spatial analysis across multiple environments. It allows you to execute ITK or arbitrary C++ code in a web browser, Node.js, Python, or via WASI runtimes.

    Key capabilities include:

    • Building C/C++ to WebAssembly: Tools to compile C++ code for Wasm targets.
    • Data Bridging: Bridging local filesystems, JavaScript/TypeScript/Python data structures, and traditional file formats.
    • Efficient Data Transfer: High-performance data movement in and out of the WebAssembly runtime.
    • Parallel Execution: Asynchronous, parallel execution of processing pipelines using a worker pool.

  3. Available @itk-wasm packages and capabilities

    main

    The following packages are built with itk-wasm and provide various scientific and medical imaging capabilities for both JavaScript/TypeScript and Python environments:

    • @itk-wasm/compare-images: Compare images with a tolerance for regression testing.
    • @itk-wasm/compress-stringify: Zstandard compression/decompression and base64 encoding/decoding in WebAssembly.
    • @itk-wasm/dicom: Read files and images related to the DICOM file format.
    • @itk-wasm/downsample: Pipelines for downsampling images.
    • @itk-wasm/elastix: A toolbox for rigid and nonrigid registration of images.
    • @itk-wasm/htj2k: Wasm-SIMD accelerated decoding and encoding of High-throughput JPEG2000 (HTJ2K) images.
    • @itk-wasm/image-io: Input and output for scientific and medical image file formats.
    • @itk-wasm/mesh-filters: Mesh filters to repair, remesh, subdivide, decimate, smooth, triangulate, etc.
    • @itk-wasm/mesh-io: Input and output for scientific and medical mesh file formats.
    • @itk-wasm/morphological-contour-interpolation: Morphological contour interpolation for scientific and medical mesh file formats.

  6. Use the TypeScript / JavaScript core tooling

    main

    The itk-wasm NPM package provides the core infrastructure for using ITK-Wasm in Node.js and browser environments. It includes:

    • Runtime Support: Works in both Node.js and web browsers.
    • Web Worker Pool: Enables parallel processing for performance.
    • CLI: A command line interface used to drive builds, generate language bindings/package configurations, and test wasm binaries.
    npm install itk-wasm

  9. Access transform parameters via the transform__parameters slot

    main

    In the ITK-Wasm model, a transform object provides access to its internal parameters through the transform__parameters slot. This slot returns the parameters of the transform as BinaryData.

    • Domain: None (no input required to access the slot)
    • Range: [1..1] [BinaryData] (returns exactly one BinaryData object representing the parameters)

  10. Understand the Mesh data model

    main

    The Mesh class is an InterfaceType used to represent an N-dimensional mesh. It serves as a structured way to pass mesh data between WebAssembly modules and the host environment. A Mesh consists of spatial coordinates (points), connectivity information (cells), and optional attribute data associated with either the points or the cells.

    Key components of a Mesh include:

    • Points: The spatial coordinates of the mesh vertices.
    • Cells: The connectivity information defining how points form cells (e.g., triangles, tetrahedrons).
    • Point Data: Attributes associated with individual points (e.g., scalars, vectors).
    • Cell Data: Attributes associated with individual cells (e.g., cell volumes, normals).