ECG_QC (Quality Classification)

Full Documentation : https://aura-healthcare.github.io/ecg_qc/

Website : https://www.aura.healthcare

Github : https://github.com/Aura-healthcare

Version : 1.0b6

ecg_qc is a python library that classifies ECG signal into good/bad quality thanks to Machine Learning.

There are currently 4 trained models:

Name	Type of model	(training) ECG time window	(training) ECG segments individual standardization
dfc_2s.pkl	Decision Tree Classifier	2 seconds	No
rfc_2s.pkl	Random Forest Classifier	2 seconds	No
rfc_2s_norm.pkl	Random Forest Classifier	2 seconds	Yes
xgb_9s.joblib	XGboost Classifier	9 seconds	No

Installation / Prerequisites

Dependencies

ecg_qc requires:

Python (>= 3.6)
biosppy>=0.6.1
dill>=0.3.4
pathtools>=0.1.2
py-ecg-detectors>=1.0.2
scikit-learn>=0.23.2
wfdb>=3.1.1
xgboost>=1.3.1

User installation

The easiest way to install hrv-analysis is using pip :

$ pip install ecg-qc

you can also clone the repository:

$ git clone https://github.com/Aura-healthcare/ecg_qc.git
$ python setup.py install

Getting started

Usage

Import:

from ecg_qc import EcgQc

Class initialization:

ecg_qc = EcgQc()

Default parameters:

ecg_qc = EcgQc(model='rfc_norm_2s.pkl',
               sampling_frequency=256,
               normalized=True)

Predicting the quality of the signal:

ecg_data = [1905.72, ... -150.75995323, -134.14559104] # ECG values with same sampling frequency as class declaration

signal_quality = ecg_qc.get_signal_quality(ecg_data)

Computing SQIs before making prediction:

ecg_data = [1905.72, ... -150.75995323, -134.14559104] # ECG values with same sampling frequency as class declaration

sqi_scores = ecg_qc.compute_sqi_score(ecg_data)
signal_quality = ecg_qc.predict_quality(sqi_scores)

Authors

Alexandre CHIROUZE - (https://github.com/achirouze)

Alexis COMTE - (https://github.com/alexisgcomte)

Laura DUMONT - (https://github.com/laudmt)

License

This project is licensed under the GNU GENERAL PUBLIC License - see the LICENSE.md file for details

References

Nemcova, A., Smisek, R., Opravilová, K., Vitek, M., Smital, L., & Maršánová, L. (2020). Brno University of Technology ECG Quality Database (BUT QDB) (version 1.0.0). PhysioNet. https://doi.org/10.13026/kah4-0w24.