Vortex

About

vortex is a high-performance C++ library with Python bindings for rapid development of real-time OCT engines.

vortex is developed and maintained by Mark Draelos of the Image-Guided Medical Robotics Lab at the University of Michigan to facilitate the prototyping of new application-specific optical coherence tomography (OCT) software. It is a general-purpose library for OCT software development that seeks to:

promote reproducibility in data acquisition and processing,
guarantee synchronization between data streams,
support on-the-fly changes for interactive use cases,

integrate cleanly with user interfaces,
facilitate rapid prototyping with high-level abstractions and sane defaults, and
minimize restrictions and assumptions.

vortex is released under the permissive BSD-3 license to facilitate its use in both research and commercial contexts.

Features

Through its modular architecture, vortex encapsulates a complete OCT engine, including scan pattern generation, data acquisition and processing, synchronized I/O, and persistent storage.

Scan Pattern Generation

Generate common scan patterns with helper classes or design your own in C++ or Python. Interleave multiple scans and route to separate storage after processing. Queue scans to the engine for execution and interrupt as needed. Insert signaling events to synchronize scan pattern changes with post-processing code.

Truncated radial scan (color) with time-optimal turnaround (white) between segments.

Real-Time OCT Processing

Use GPU or FPGA acceleration for real-time OCT processing. Spread large workloads across multiple GPUs. Tune block size and stream parallelism to your use case. Perform background subtraction, resampling, and filtering of raw spectra.

Real-time OCT processing and display while scanning a 3D printed vortex logo.

Extensive Python Bindings

Create real-time OCT engines exclusively using Python. Transparently share data with vortex using NumPy and CuPy. Lock shared memory for concurrent access from Python and C++. Asynchronously chain custom GPU processing onto the C++ OCT pipeline using CuPy.

pc = CUDAProcessorConfig()

# window + dispersion correction
window = np.hanning(pc.samples_per_ascan)
phasor = dispersion_phasor(len(window), \
    cfg.dispersion)
pc.spectral_filter = window * phasor

# DC subtraction per block
pc.average_window = pc.ascans_per_block

p = CUDAProcessor()
p.initialize(pc)

Python code for configuring OCT processing with GPU acceleration.

Synchronized I/O

Drive external hardware with scan pattern waveforms. Compensate for I/O delays with scan pattern lookahead. Add additional channels for multi-scanner hardware. Generate scan-derived digital pulses to trigger other devices.

Segment start pulses (yellow) for a radial scan (blue) with variable-length turnaround using the volume start trigger.

Flexible Data Storage

Stream spectra or processed data to disk in common or raw file formats. Buffer high-speed acquisitions in memory as needed. Reassemble out-of-order scan patterns transparently as they write to disk.

vortex natively supports NumPy, HDF5^®, and MATLAB^® file formats.

UI Interactivity

Create interactive applications using callbacks from vortex's multithreaded engine. Display live OCT data as it is acquired or modify an ongoing acquisition.

Example user interface in Python for live OCT display using vortex.

Users

Both academic and industry entities use vortex due to its permissive BSD-3 license.

Diagnostic Imaging & Image-Guided Interventions Laboratory

Rollins Biophotonics
Lab @ CWRU

Digital Eye Lab @ Duke

The following publications use vortex in their scanning, acquisition, and/or processing.

Mark Draelos, “Time-Optimal Spiral Trajectories with Closed-Form Solutions,” IEEE Robotics and Automation Letters, vol. 8, no. 4, pp. 2213-2220, April 2023.
Mark Draelos, Christian Viehland, Ryan P. McNabb, Anthony N. Kuo, and Joseph A. Izatt, “Adaptive point-scan imaging beyond the frame rate-resolution limit with scene-reactive scan trajectories,” Optica, vol. 9, no. 11, pp. 1276-1288, Nov. 2022.
Jingkai Zhang, Ruobing Qian, Kevin C. Zhou, Christian Viehland, Mark Draelos, Al-Hafeez Dhalla, and Joseph A. Izatt, “Real-time wide-field spectral-scanning FMCW coherent 3D imaging and velocimetry,” Optics Letters 47, 4064-4067 (2022).
Pablo Ortiz, Mark Draelos, Amit Narawane, Ryan P. McNabb, Anthony N. Kuo and Joseph A. Izatt, “Robotically-Aligned Optical Coherence Tomography with Gaze Tracking for Live Image Montaging of the Retina,” IEEE International Conference on Robotics and Automation (ICRA), 2022, pp. 3783-3789.
Pablo Ortiz, Mark Draelos, Christian Viehland, Ruobing Qian, Ryan P. McNabb, Anthony N. Kuo, and Joseph A. Izatt, “Robotically aligned optical coherence tomography with 5 degree of freedom eye tracking for subject motion and gaze compensation,” Biomedical Optics Express, vol. 12, no. 12, pp. 7361-7376, Dec. 2021

Releases

vortex is under active development with major releases and documentation listed below.

C++ users may download the source below and follow the Build Guide. Python users may install pre-built binary wheels for standard hardware configurations using the package installer for Python, taking care to choose the correct CUDA version. See Getting Started for more details.

pip install vortex-oct-cuda112 vortex-oct-tools -f https://www.vortex-oct.dev/

Version	Status	C++ Source	Python Binary Wheels			Links
Version	Status	C++ Source	CUDA 10.2	CUDA 11.2	CUDA 11.6	Links
v0.4.3 6/30/2022			3.6 3.7 3.8 3.9 3.10	3.6 3.7 3.8 3.9 3.10	3.7 3.8 3.9 3.10	Documentation Release Notes
v0.4.2 3/3/2022			3.6 3.7 3.8 3.9 3.10	3.6 3.7 3.8 3.9 3.10		Documentation Release Notes
v0.4.1 1/22/2022			3.6 3.7 3.8 3.9 3.10	3.6 3.7 3.8 3.9 3.10		Documentation Release Notes

= latest release for starting new projects

= latest maintenance release to support existing projects

Citation

Please use the following citation for vortex if you found it useful in your research.

@misc{Vortex,
    author = {Mark Draelos},
    year   = {2022},
    url    = {https://www.vortex-oct.dev/},
    title  = {Vortex -- An open-source library for building real-time OCT engines in C++ or Python}
}

Download the citation in BibTex or RIS format. Check back later as this citation will likely change in the future.