electrography

This Python package is intended for use with OpenBCI boards and/or their data. You can use it to stream data from any of the OpenBCI boards, add triggers to data files, and to analyse the results.

Dependencies

• BrainFlow
• NumPy
• SciPy
• scikit-learn

Citation

A tutorial paper associated with this package will be made available soon. Please cite this if you use electrography in your work.

• Citation to follow!

Example Data Acquisition

import time
from electrography.boards import OpenBCI

# Create a new OpenBCI instance, in this example for a Ganglion board.
# Note that the `serial_port` address may be different on your system!
board = OpenBCI("ganglion", serial_port="/dev/ttyACM0", \
    log_file="example_egg.tsv")

# Stream for 30 seconds, sending a trigger after 1 second.
board.start_stream()
time.sleep(1)
board.trigger(200)
time.sleep(29)
board.stop_stream()

# Close the connection
board.close()

Example Data Analysis

from openbci.data import read_file, get_physiology, hampel, \
    butter_bandpass_filter, movement_filter

# Set the low (high-pass) and high (low-pass) bounds for the data (EGG). For 
# electrogastrography, you'd typically use 0.5-10 cycles per minute (cpm), i.e.
# 0.0083-0.17 Herz (Hz).
low_bound = 0.0083
high_bound = 0.17

# Define the normogastric band (2-4 cpm, i.e. 0.033-0.067 Hz) and main 
# frequency (3 cpm, i.e. 0.05 Hz).
normogastric_freq = 0.05
normogastric_band = [0.033, 0.067]

# Load the data from a file at `fpath`.
raw = read_file(fpath)
data, t, triggers, sampling_rate = get_physiology(raw, "ganglion")

# Mean-centre the data.
m_per_channel = numpy.nanmean(data, axis=1, keepdims=True)
data = data - m_per_channel

# Hampel-filter the data.
data = hampel_filter(data, k=2000, n_sigma=3)

# Filter the data for EGG frequencies.
for channel in range(data.shape[0]):
    data[channel,:] = butter_bandpass_filter(data[channel,:], \
        low_bound, high_bound, sampling_rate, bidirectional=True)

# Employ a movement filter for frequencies below the normogastric frequency.
data, noise = movement_filter(data, sampling_rate, normogastric_freq)

# Perform ICA denoising on the data, using the normogastric band.
data = ica_denoise(data, sampling_rate, low_bound, high_bound, \
    normogastric_band, snr_threshold=3.0)