mach: ultrafast ultrasound beamforming
Open-source GPU tool processes data 10x faster than existing solutions, running on consumer hardware.
A research team from the University of North Carolina and Stanford has released 'mach', a breakthrough open-source ultrasound beamforming library that shatters previous computational barriers. The tool leverages a highly optimized CUDA kernel and a hybrid delay computation strategy to process volumetric ultrasound data at 1.1 trillion points per second on a consumer-grade GPU. This represents a >10x speedup over existing open-source GPU beamformers and completes benchmark reconstructions in just 0.23 ms—six times faster than the acoustic round-trip time to the imaging depth itself. Crucially, it maintains numerical accuracy with errors below -60 dB for Power Doppler and -120 dB for B-mode imaging.
By eliminating the beamforming bottleneck, mach enables real-time applications previously impossible on consumer hardware. This includes 3D functional neuroimaging, intraoperative surgical guidance, and ultrasound localization microscopy—emerging modalities that require processing massive datasets from high-frame-rate, dense reconstruction grids. The accessible Python interface and open-source availability mean researchers can immediately apply this tool to accelerate medical physics research and develop new clinical applications. The work has been peer-reviewed and published in the SPIE Journal of Medical Imaging, with source code freely available for the community.
- Processes 1.1 trillion points per second on consumer GPUs, achieving >10x speedup over existing solutions
- Completes the PyMUST rotating disk benchmark reconstruction in 0.23 ms (6x faster than acoustic round-trip time)
- Open-source Python/CUDA tool enables real-time 3D ultrafast ultrasound for neuroimaging and surgery for the first time
Why It Matters
Democratizes advanced medical imaging by making real-time 3D ultrasound possible on affordable hardware, accelerating research and clinical adoption.