Skip to content

Latest commit

 

History

History
92 lines (66 loc) · 5.55 KB

README.md

File metadata and controls

92 lines (66 loc) · 5.55 KB
Raw

BrainOn: an online brain-computer interface (BCI) framework for feature modulation and processing

Welcome!

Hello！First and foremost, thank you for taking your time to visit the BrainOn repository. Hope it can help or inspire you.

This BrainOn project, which was developed in python with multi-threading concurrency programming, is aimed to create an online brain-computer interface (BCI) framework for feature modulation and processing, allowing researchers to develop their own online experiment programs easily and quickly. Notably, only method recognition of class BaseProcessingRecog needs to be overridden by the user to implement the online functionality. The BrainOn supports real-time data stream reading, data processing (e.g., re-reference, down-sampling, filter, data segmentation, and so on), real-time result feedback,pattern recognition (e.g., machine learning, deep learning), as well as other functions. Moreover, The BrainOn can be used to control devices, such as classic brain-controlled typing (see https://www.youtube.com/watch?v=EW2Q08oHSBo), brain-controlled robotic arms (see https://www.youtube.com/watch?v=A1w-e2dBGl0).

The BrainOn is a part of MetaBCI that is meant to provide a python platform for BCI users to design paradigm, collect data, process signals, present feedbacks and drive robots. Please see https://github.com/TBC-TJU/brainda for more information on Brainda, which is the other part of MetaBCI.

This document is to give you more information about the BrainOn. Click the section name to jump to corresponding location.

Overview

Filelist

  • BaseFramework.py
    A BCI online baseframework with multi-thread concurrency programming was proposed, which included three modules, i.e. class BaseReadData for reading data from EEG device in real time., class BaseProcessingRecog for signal processing and pattern recognition, class SendMessageUdp for message transmission (e.g., real-time classification result).

  • BasePreProcessing.py
    A preprocessing baseframework is mainly used for building offline model for the further online experiment with the training problems.

  • ReadNeuroscan.py
    A demo of reading live data streaming from the Neuroscan Synamps2 amplifier (a EEG acquisition device, Compumedics Ltd., Melbourne Australia) was implemented by deriving from parent class BaseFramework. Although the demo is based on Neuroscan device, it's worth noting that the framework is theoretically compatible for most of EEG devices.

  • simulate_serve_new_packet.py
    Simulate EEG amplifier acting as a server and Distribute data in the fixed time interval. The software USR-TCP232-TEST can be used as a client to debug. The download link is https://www.pusr.com/support/downloads/usr-tcp232-test-V13.

  • simulate_online_CNT.py
    Simulate the online processing flow for program validity that not need connect amplifier or bulid hardware platform. It is convenient for the program validation of the new online experiment without the need for a hardware platform.

  • demo_offline_model.py
    A demo of building offline model was developed based on the hybrid P300-SSVEP BCI by deriving from parent class BasePreProcessing.

  • demo_online_hybridBCI.py
    A demo of BCI online modulation and processing framework was implemented based on the hybrid P300-SSVEP BCI system using above base frameworks. Notably, this program can run properly if only the hardware platform is connected, which including an EEG amplifier (Neuroscan in this case), ethernet network cable, and stimulation program.

  • algorithms-folder
    Some BCI algorithms to complement Brainda repository, such as Spatial-Temporal Discriminant Analysis (STDA), Shrinkage Linear discriminant analysis (SKLDA), Hierarchical discriminant component analysis (HDCA), Sliding-HDCA, Discriminative Canonical Pattern Matching (DCPM), and so on.
    The interface form is consistent with scikit-learn for ease of use.

Flow Diagram

The BrainOn is a three-threads online processing framework based on the concurrency programming, where the main thread is mainly used for controlling the overall experimental process, the processing thread is mainly used for data preprocessing, pattern recognition, result communication, and the read thread is mainly used for reading real-time data streaming from the EEG amplifier. The flow diagram is shown below.

Contributor

The project was developed and completed in the Lab of Neural Engineering and Rehabilitation, Tianjin University, China.

Dr.Jin Han is the main contributor to the BrainOn project.

Dependency

  • joblib>=1.0.1
  • mne==0.23.4
  • numpy==1.21.3
  • python>=3.8
  • scikit-learn==1.0.2
  • scipy==1.7.1

Abbreviation

  • BCI: brain-computer interface
  • EEG: electroencephalogram
  • P300: P300 waveform
  • SSVEP: steady-state visual evoked potential
  • MI: motor imagery

Troubleshooting

Please feel free to contact us ( jinhan9165@gmail.com ) if there are any questions.