EyeLinkIO 👀

A lightweight library to import SR Research EDF files into Python.

This Software is currenly pre-alpha, meaning it is currently being developed: Changes to the API (function names, etc.) may occur without warning. This library has been tested with MacOS and Linux, but not Windows.

About the Eyelink Data Format

The EyeLink Data Format (EDF; not to be confused with the European Data Format) is used for storing eyetracking data from EyeLink eyetrackers. It was put forward by the company SR Research. SR Research EDF files store data in a binary format, and reading these files currently requires the eyelink-edfapi C library that is included in the EyeLink Software Development Kit.

Dependencies

Strictly speaking, EyeLinkIO only requires Numpy. For converting data to pandas DataFrames or MNE-Python Raw instances, you must have those respective packages installed.

Important

• You must have the EyeLink Software Development Kit installed on your computer
• You must register an account on the forum to access the download (registration is free)

Installation

1. Stable Installation

pip install eyelinkio

2. Development Installation (For those who need features or bugfixes that aren't released yet):

pip install git+https://github.com/scott-huberty/eyelinkio.git

3. Editable Installation (For contributors to EyeLinkIO):

pip install -e ./eyelinkio

Important

• Fork the repository on GitHub first.
• Clone your forked repository to your local machine.
• Make sure you're in the directory containing the cloned eyelinkio folder when you run the command above

This package is not currently available on Conda.

Example Usage

Reading an EDF file

from eyelinkio import read_edf
eyelinkio.utils import _get_test_fnames  # for demonstration purposes only

fname = _get_test_fnames()[0]  # Replace this function with the path to your EDF file
edf_file = read_edf(fname)
print(edf_file)

<EDF | test_raw.edf> 
  Version: EYELINK II 1 
  Eye: LEFT_EYE 
  Pupil unit: PUPIL_AREA 
  Sampling frequency: 1000.0 Hz 
  Calibrations: 1 
  Length: 66.827 seconds 

Inspecting an EDF object

edf_file.keys()

Out: dict_keys(['info', 'discrete', 'times', 'samples'])

edf_file["info"].keys()

Out: dict_keys(['meas_date', 'version', 'camera', 'serial', 'camera_config', 'sfreq', 'ps_units', 'eye', 'sample_fields', 'edfapi_version', 'screen_coords', 'calibrations', 'filename'])

edf_file["discrete"].keys()

Out: dict_keys(['messages', 'buttons', 'inputs', 'blinks', 'saccades', 'fixations'])

Exporting an EDF object to Pandas or MNE-Python

# Convert to a pandas DataFrame or an MNE Raw instance
dfs = edf_file.to_pandas()
raw, calibration = edf_file.to_mne()

See the documentation for more.

Acknowledgements

This package was originally adapted from the pyeparse package (created by several of the core developers of MNE-Python). It copies much of the EDF (Eyelink Data Format) reading code.

Limitations

• Reading extra sample fields (velocity, HREF, head position etc.) from the EDF file is not yet supported.

See the Roadmap for more details.