SP-BAND

Implementation of Spectral Parameterization for the Broadband Analysis of Neural Data (SP-BAND), a work presented at the Asilomar Conference on Signals, Systems, and Computers on October 28, 2024.

This work is an extension of the spectral parameterization algorithm proposed by Donoghue et al. (2020) "Parameterizing neural power spectra into periodic and aperiodic components``. Nature Neuroscience, 23, 1655-1665. DOI: 10.1038/s41593-020-00744-x.

The ParamSpectra class is built off the FOOOF class developed by Donoghue et al. which is available at fooof-tools.github.io. The work in this SP-BAND package is not affiliated with the FOOOF package.

This work offers a data-driven approach to spectral parameterization that incorporates prior knowledge of expected periodic activity. As a result, this improves fit to real-world broadband data and improves the interpretability of the parameterized components.

Installation

To use the SP-BAND package, first clone the repository:

git clone [email protected]:aroyphillips/SP-BAND.git

Then, navigate to the directory and install the conda environment:

cd SP-BAND
conda env create -f environment.yml

This will create the spband environment. To activate the environment, run:

conda activate spband

Finally, install the package:

pip install -e .

Now, you are ready to use the SP-BAND package!

Usage

To use the SP-BAND, visit the notebooks directory for a demonstration of the package.

Repository Structure

The repository is structured as follows:

SP-BAND/
│
├── notebooks/  # Jupyter notebooks demonstrating the SP-BAND package
|   ├── 1.0.0-SP-BAND-Demonstration.ipynb
│
├── SPBAND/  # Source code for the SP-BAND package
│   ├── __init__.py
│   ├── spband.py
│   ├── utils.py
│
|-- data/  # Data used in the notebooks
│
|-- environment.yml  # Conda environment file
|
|-- LICENSE  # MIT License
|
|-- project.toml  # Project file
│
|-- README.md  # This README file