py_neuromodulation

Analyzing neural data can be a troublesome, trial and error prone, and beginner unfriendly process. py_neuromodulation allows using a simple interface for extraction of established neurophysiological features and includes commonly applied pre -and postprocessing methods.

Only time series data with a corresponding sampling frequency are required for feature extraction.

The output will be a pandas.DataFrame <https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.html>_ including different time-resolved computed features. Internally a stream get's initialized, which resembles an online data-stream that can in theory also be be used with a hardware acquisition system.

The following features are currently included:

• oscillatory: fft, stft or bandpass filtered band power
• temporal waveform shape <https://www.sciencedirect.com/science/article/pii/S1364661316302182>_
• fooof <https://fooof-tools.github.io/fooof/>_
• mne_connectivity estimates <https://mne.tools/mne-connectivity/stable/index.html>_
• Hjorth parameter <https://en.wikipedia.org/wiki/Hjorth_parameters>_
• non-linear dynamical estimates <https://nolds.readthedocs.io/en/latest/>_
• various burst features
• line length
• and more...

Find here the preprint of py_neuromodulation called "Invasive neurophysiology and whole brain connectomics for neural decoding in patients with brain implants" [1]_.

The original intention for writing this toolbox was movement decoding from invasive brain signals [2]_. The application however could be any neural decoding problem. py_neuromodulation offers wrappers around common practice machine learning methods for efficient analysis.

Find the documentation here http://py-neuromodulation.readthedocs.io for example usage and parametrization.

Installation

py_neuromodulation requires at least python 3.10. For installation you can use pip:

.. code-block::

pip install py-neuromodulation

Alternatively you can also install the package in a conda environment:

.. code-block::

conda create -n pynm-test python=3.11
conda activate pynm-test

Then install the packages listed in the pyproject.toml:

.. code-block::

pip install .

Optionally the ipython kernel can be specified for the installed pynm-test conda environment:

.. code-block::

ipython kernel install --user --name=pynm-test

Then py_neuromodulation can be imported via:

.. code-block::

import py_neuromodulation as nm

Basic Usage

.. code-block:: python

import py_neuromodulation as nm
import numpy as np

NUM_CHANNELS = 5
NUM_DATA = 10000
sfreq = 1000  # Hz
sampling_rate_features_hz = 3  # Hz

data = np.random.random([NUM_CHANNELS, NUM_DATA])

stream = nm.Stream(sfreq=sfreq, data=data, sampling_rate_features_hz=sampling_rate_features_hz)
features = stream.run()

Check the Usage <https://py-neuromodulation.readthedocs.io/en/latest/usage.html>_ and First examples <https://py-neuromodulation.readthedocs.io/en/latest/auto_examples/plot_0_first_demo.html>_ for further introduction.

Contact information

For any question or suggestion please find my contact information at my GitHub profile <https://github.com/timonmerk>_.

References

.. [1] Merk, T. et al. Invasive neurophysiology and whole brain connectomics for neural decoding in patients with brain implants, https://doi.org/10.21203/rs.3.rs-3212709/v1 (2023). .. [2] Merk, T. et al. Electrocorticography is superior to subthalamic local field potentials for movement decoding in Parkinson’s disease. Elife 11, e75126, https://doi.org/10.7554/eLife.75126 (2022).