ecgmentations 0.0.9

Ecg augmentation library and easy to use wrapper around other libraries

Ecgmentations

Ecgmentations is a Python library for ecg augmentation. Ecg augmentation is used in deep learning to increase the quality of trained models. The purpose of ecg augmentation is to create new training samples from the existing data.

Here is an example of how you can apply some augmentations from Ecgmentations to create new ecgs from the original one:

Table of contents

• Authors
• Installation
• A simple example
• List of augmentations
• Citing

Authors

Rostislav Epifanov — Researcher in Novosibirsk

Installation

Installation from PyPI:

pip install ecgmentations

Installation from GitHub:

pip install git+https://github.com/rostepifanov/ecgmentations

A simple example

import numpy as np
import ecgmentations as E

# Declare an augmentation pipeline
transform = E.Sequential([
    E.TimeReverse(p=0.5),
    E.ChannelShuffle(p=0.06),
])

# Create example ecg (length, nchannels)
ecg = np.ones((5000, 12))

# Augment an ecg
transformed = transform(ecg=ecg)
transformed_ecg = transformed['ecg']

List of augmentations

The list of time axis transforms:

• TimeReverse
• TimeShift
• TimeSegmentShuffle
• RandomTimeWrap
• TimeCutout
• TimeCrop
• CenterTimeCrop
• RandomTimeCrop
• TimePadIfNeeded
• Pooling
• Blur

The list of pulse transforms:

• SinePulse
• PowerlineNoise
• RespirationNoise
• SquarePulse

The list of filter transforms:

• LowPassFilter
• HighPassFilter
• BandPassFilter
• SigmoidCompression

The list of other transforms:

• AmplitudeInvert
• ChannelShuffle
• ChannelDropout
• GaussNoise
• GaussBlur
• AmplitudeScale

Citing

If you find this library useful for your research, please consider citing:

@misc{epifanov2023ecgmentations,
  Author = {Rostislav Epifanov},
  Title = {Ecgmentations},
  Year = {2023},
  Publisher = {GitHub},
  Journal = {GitHub repository},
  Howpublished = {\url{https://github.com/rostepifanov/ecgmentations}}
}