PhysioNet Cardiovascular Signal Toolbox

Introduction

The PhysioNet Cardiovascular Signal Toolbox is an open-source modular program for calculating heart rate variability (HRV) implemented in Matlab with evidence-based algorithms and output formats. The Toolbox is compatible with 64-bit MATLAB on GNU/Linux, Mac OS X, and MS-Windows.

• It was compared to several other open source and proprietary tools including the PhysioNet C HRV Toolkit to create a benchmark for the field.
• It was shown to be equivalent to the PhysioNet C HRV Toolkit
• It contains the most extensive set of tools in any HRV algorithm collection so far published. (See list below )
• It has no dependencies outside of Matlab (tested on Matlab R2017a and R2017b).

Despite its popularity in research and relatively long history, there is still much disagreement in the methods by which researchers apply HRV signal processing. This disagreement limits meaningful comparisons between studies and scientific repeatability, especially when in-house, custom, non-public software are used.

The PhysioNet Cardiovascular Signal Toolbox has been developed to address the issues of validation, standardization, and repeatability. It has been designed to accept a wide range of cardiovascular signals and analyze those signals with a variety of classic and modern signal processing methods. The toolbox includes many features not offered in other programs, including peak and pulse detection, signal quality analysis, rhythm detection, beat classification, general HRV statistics, phase rectified signal averaging (PRSA) techniques for deceleration and acceleration capacity, Detrended Fluctuation Analysis (DFA), Heart Rate Turbulence (HRT), Multiscale Entropy (MSE).

The toolbox can process raw waveform data (such as electrocardiograms) as well as derived RR-interval data. Although it was designed not to deal with file formats, the toolbox natively supports MAT, CSV, or WFDB-compatible annotation formats without relying on PhysioNet’s WFDB libraries (or other libraries). If users wish to export results from the HRV Toolbox, a function is included that allows for standard WFDB compatible output annotation files or CSV output files.

The PhysioNet Cardiovascular Signal Toolbox employs several methods to prepare data for HRV estimation, including assessing signal quality and detecting arrhythmias, erroneous data, and noise. These segments of data, which must be excluded from HRV analysis, can then be systematically removed based on threshold settings selected by the user or recommended in previously validated studies. In particular, our toolbox contains one initialization file which lists all the options available, with typical default settings. In this way, a user may easily identify which settings need to be given considerable thought (all the ones listed) and provide this listing in a publication.

The Toolbox is open-source (distributed under the GNU GPL (v3)). Sources for the current version of the Toolbox are available here (signature). For development snapshots, see the project repository on GitHub.

Key Contributions

You might ask, why *another* HRV toolbox? The following is a list of key contributions this toolbox and accompanying publication makes to the field, and why you might want to use this in preference to other toolboxes and software out there.

• A full suite of waveform processing tools, for end-to-end processing,
• Detailed explanations of preprocessing and parameter choices to identify where divergences in methods can occur, and to provide standardization in the field,
• Benchmarked against other open source HRV tools to identify when they disagree with each other,
• Benchmarked against PhysioNet's C code for compatibility, and hence can be used as a prototyping platform before switching to C for large scalable tasks or embedded systems,
• A mostly comprehensive collection of standard and more recent HRV tools that go beyond other toolboxes,
• Fully scriptable with no libraries outside Matlab required for reading data and annotations,
• Fully WFDB-compatible,
• Compatibility with commercial software often used by clinicians (e.g. Kubios),
• Open source and versioned on Github so the community may build upon it.

Quick Start

Prerequisites:

• 64-bit GNU/Linux, Mac OS X 10.9, or MS-Windows
• MATLAB R2017b or later, with Signal Processing Toolbox, Statistics and Machine Learning Toolbox, and Neural Network toolbox.

How to Use the PhysioNet Cardiovascular Signal Toolbox:

1. Download. Install and Open MATLAB.
2. Add the PhysioNet Cardiovascular Signal Toolbox folder and subfolders to your Matlab path
3. Test the Toolbox:

   For a demonstration of the toolbox, go into the Demos subdirectory and run one of the available demonstrations:

   DemoAnnotatedData

   If these demos do not run successfully, please see the Toolbox FAQ for troubleshooting hints.

Toolbox Functions

The PhysioNet Cardiovascular Signal Toolbox utilizes a standardized approach to preprocess data and compute HRV metrics using Matlab functions.

Frequently Asked Questions

Please make sure you check our list of frequently asked questions before contacting us! For the list of frequently asked questions, see our FAQ.

For questions, contributions or feedback, please post on our GitHub page:

https://github.com/cliffordlab/PhysioNet-Cardiovascular-Signal-Toolbox/issues

Acknowledgments

The PhysioNet Cardiovascular Signal Toolbox is a collection of algorithms designed and created over the last 20 years by Gari Clifford, his students and postdocs, and other collaborators, dilligently assembled, stress tested, updated, documented and Adriana N. Vest and Giulia Da Poian. Original contributors of open source code are credited in their respective MATLAB functions.

We would particularly like to thank the following people for contributing their code: Qiao Li, Patrick McSharry, Shamim Nemati, James Sun. We would also like to thank Mika Tarvainen, Raphael Schnieder, Joe Mietus, George Moody and Danny Kaplan for providing (and running) source code for comparisons, benchmarking, and stress testing.

Thanks are also due to Amit Shah, Roger Mark, Ary Goldberger for providing clinical insights during the process of creation.

Previous Releases

Previous releases of the PhysioNet Cardiovascular Signal Toolbox can be found here!.