ECG-Kit

Software Description

The main feature of this toolbox is that it allows the use of several popular algorithms for ECG processing, including:

• Algorithms from Physionet’s WFDB software package
• QRS detectors, such as gqrs, wqrs, wavedet, ecgpuwave, Pan & Tompkins, EP limited
• Wavelet-based ECG delineator
• Pulse wave detectors as wabp and wavePPG
• a2hbc and EP limited heartbeat classifiers

The toolbox also includes scripts for inspecting, correcting and reporting results from these algorithms.

ECG-kit has a common application programmer interface (API) implemented in Matlab under Windows, Linux or Mac. The kit also implements a recording interface that allows processing several ECG formats, such as MIT, ISHNE, HES, Mortara, and AHA, of arbitrary recording size. (The record so far is a one-week recording of 3 leads, sampled at 500 Hz).

The project website is http://marianux.github.io/ecg-kit/. The directory listing below provides links to components of a stable version (1.0).

Below are links to

• Installation documentation.
• A YouTube video.
• A GitHub code repository for issue tracking and continuous update.

recording208.png shows a report of the mitdb/208 recording created with the ecg-kit toolbox. The QRS detections created with several algorithms are shown in different colours (vertical dotted lines ended with triangles). In addition, the ecg delineation is represented as coloured boxes superimposed to the signal. Finally the heartbeat classification is printed above each heartbeat.

Sample Input

The documentation includes a first example and another example, as well as tutorials and further documentation.

Data from Physionet are included with the kit in the recordings subdirectory. ECG-kit has been tested on Windows 7 and Linux Ubuntu and Debian platforms.

Acknowledgments

Many thanks to Andrés Demski from UTN who helped to this project before he learned how to use it. Thanks also to all the friends in Zaragoza, Porto and Lund, and especially to the ones closest to the project:

• Pablo Laguna, Juan Pablo Martínez, Rute Almeida and Juan Bolea, for the wavedet ECG delineator and many parts of the Biosig browser project that were adapted to this project.
• Jesús Lázaro and Eduardo Gil for the PPG/ABP pulse detection code.

We also acknowledge all those listed below, who were important in many ways to the fulfilment of this project:

• George Moody, Wei Zong, Ikaro Silva, for all the software of Physionet
• Reza Sameni, for his Open-Source ECG Toolbox (OSET)
• Bob Duin and all the team behind PRtools
• Yair Altman from undocumented Matlab