FAST-Imaging/FAST

A framework for high-performance medical image processing, neural network inference and visualization

What it solves

FAST is designed to simplify the high-performance processing, neural network inference, and visualization of medical images. It removes the complexity of manual memory management across different processors (CPUs and GPUs) and provides a unified interface for deploying deep learning models in real-time medical imaging pipelines.

How it works

The framework uses modern C++, OpenCL, and OpenGL to leverage multi-core CPUs and GPUs. It provides a common interface for neural network inference using backends like TensorRT, OpenVINO, TensorFlow, and ONNX Runtime, supporting various model formats including ONNX and protobuf. It manages data coherence across storage areas automatically, allowing developers to focus on the algorithmic logic rather than explicit memory handling.

Who it’s for

Researchers and developers working with medical imaging, specifically those needing to process 2D/3D images, surface meshes, and temporal data in real-time using deep learning and high-performance computing.

Highlights

  • Unified Inference Interface: Supports multiple backends (TensorRT, OpenVINO, TensorFlow, ONNX Runtime) and formats (ONNX, protobuf, SavedModel).
  • Data Streaming: Handles static and dynamic data from sources like webcams, Intel RealSense, and ultrasound scanners.
  • High-Performance Algorithms: Includes parallel OpenCL implementations for Gaussian smoothing, non-local means, and marching cubes surface extraction.
  • Fast Visualization: Separate threads for rendering and computation to ensure smooth 2D and 3D visualization (volume rendering, mesh, and WSI pyramids).
  • Broad Format Support: Compatible with DICOM, NIFTI, HDF5, and whole slide images (WSI).
  • Interoperability: Available for use with Python and integrable into Qt applications.

Related

  • Project
    openvinotoolkit/openvino

    An open-source toolkit for optimizing and deploying deep learning models across CPU, GPU, and NPU hardware to boost inference performance.

  • Project
    nndeploy/nndeploy

    A high-performance AI deployment framework that uses visual workflows to deploy AI models across desktop, mobile, and edge devices.

  • Project
    jamjamjon/usls

    A cross-platform Rust library powered by ONNX Runtime for efficient inference of vision and vision-language models under 1B parameters.

  • Project
    pykeio/ort

    A Rust interface for hardware-accelerated inference and training of ONNX machine learning models, enabling efficient deployment across various hardware and platforms.

  • Project
    microsoft/onnxruntime

    A cross-platform machine learning accelerator that optimizes inference and training for models from various frameworks like PyTorch and TensorFlow.