pwstegman / Bci.js

Brain Computer Interfaces (BCIs) with JavaScript

Getting Started

Latest release is v1.8.0. You can view the release notes at releases

Documentation is available at https://bci.js.org/docs/

Node.js

npm install bcijs

Browser

<script src="https://cdn.jsdelivr.net/npm/[email protected]/dist/bci.min.js"></script>

Feature Overview

For a complete list of methods, see the docs.

Tutorials

• EEG Motor Imagery Classification in Node.js with BCI.js

Examples

More examples can be found in the examples directory

Bandpower

const bci = require('bcijs');

// Generate 1 second of sample data at 512 Hz
// Contains 8 μV / 8 Hz and 4 μV / 17 Hz
let samplerate = 512;
let signal = bci.generateSignal([8, 4], [8, 17], samplerate, 1);

// Compute relative power in each frequency band
let bandpowers = bci.bandpower(signal, samplerate, ['alpha', 'beta'], {relative: true});

console.log(bandpowers); // [ 0.6661457715567836, 0.199999684787573 ]

Epoch data

let samples = [[1,2], [3,4], ...] // 2D array where rows are samples and columns are channels
let samplerate = 256; // 256 Hz

// Epoch data into epochs of 256 samples with a step of 64 (75% overlap)
// Then find the average alpha and beta powers in each epoch.
let powers = bci.windowApply(
	samples,
	epoch => bci.bandpower(epoch, samplerate, ['alpha', 'beta'], {average: true}),
	256,
	64
);

Subscript

const bci = require('bcijs');

// 5 samples of data from 3 channels 
let signal = [[1,2,3], [5,3,4], [4,5,6], [7,5,8], [4,4,2]];

// Select the first 3 samples from channels 1 and 3
let subset = bci.subscript(signal, '1:3', '1 3'); // [ [ 1, 3 ], [ 5, 4 ], [ 4, 6 ] ]

Linear discriminant analysis

const bci = require('bcijs');

// Training set
let class1 = [[0, 0], [1, 2], [2, 2], [1.5, 0.5]];
let class2 = [[8, 8], [9, 10], [7, 8], [9, 9]];

// Testing set
let unknownPoints = [[-1, 0], [1.5, 2],	[7, 9], [10, 12]];

// Learn an LDA classifier
let ldaParams = bci.ldaLearn(class1, class2);

// Test classifier
let predictions = bci.ldaClassify(ldaParams, unknownPoints);

console.log(predictions); // [ 0, 0, 1, 1 ]

Check out https://bci.js.org/examples/lda for a visual demo of how LDA works

Usage in the web

BCI.js can be loaded from the jsDelivr CDN with

<script src="https://cdn.jsdelivr.net/npm/[email protected]/dist/bci.min.js"></script>

You can also find bci.js and bci.min.js at releases.

BCI.js methods are accessible via the global object bci.

If building a web distributable using a tool such as browserify or webpack, require bcijs/browser.js to load only methods that are browser compatible. Node.js specific methods such as networking and file system methods will not be included.

const bci = require('bcijs/browser.js');

Requiring specific methods

You can require specific methods as well. For example, if you only need fastICA, you can use

const fastICA = require('bcijs/lib/math/fastICA.js');

BCI.js methods can be found in the src/ directory.

Files are transpiled from ES6 import/export (in src/) to CommonJS (generated lib/) on npm install.

Documentation

Documentation can be found at https://bci.js.org/docs or by viewing api.md

Deprecated methods can be found at deprecated.md

Building

See dev.md for info on how to modify and build BCI.js

Reference

BCI.js began as WebBCI, a library developed to aid in my research at the Human Technology Interaction Lab at the University of Alabama Department of Computer Science. If you use BCI.js in a published work, please reference this paper

P. Stegman, C. Crawford, and J. Gray, "WebBCI: An Electroencephalography Toolkit Built on Modern Web Technologies," in Augmented Cognition: Intelligent Technologies, 2018, pp. 212–221.

Logo uses icon from Font Awesome.

Contact

If you have a commercial use case for BCI.js and would like to discuss working together, contact me at [email protected]