Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Motor Imagery Classification for Asynchronous EEG-Based Brain-Computer Interfaces

Huanyu Wu, Siyang Li, Dongrui Wu

TL;DR

Addresses asynchronous MI-BCIs that must detect MI without triggers. Proposes SWPC, a two-module system with a prescreening stage and a multi-class MI classification stage, both trained with supervised learning followed by self-supervised learning for feature refinement; test data are ingested as sliding windows of length $L_w$ and gated by threshold $\tau$. Across four BNCI-Horizon EEG datasets, SWPC achieves the highest average accuracy on both within-subject and cross-subject tasks and outperforms the best baselines by about 2 percentage points. This approach improves practicality of asynchronous BCIs and points to future work in transfer learning, test-time adaptation, and extending SWPC to other BCI paradigms.

Abstract

Motor imagery (MI) based brain-computer interfaces (BCIs) enable the direct control of external devices through the imagined movements of various body parts. Unlike previous systems that used fixed-length EEG trials for MI decoding, asynchronous BCIs aim to detect the user's MI without explicit triggers. They are challenging to implement, because the algorithm needs to first distinguish between resting-states and MI trials, and then classify the MI trials into the correct task, all without any triggers. This paper proposes a sliding window prescreening and classification (SWPC) approach for MI-based asynchronous BCIs, which consists of two modules: a prescreening module to screen MI trials out of the resting-state, and a classification module for MI classification. Both modules are trained with supervised learning followed by self-supervised learning, which refines the feature extractors. Within-subject and cross-subject asynchronous MI classifications on four different EEG datasets validated the effectiveness of SWPC, i.e., it always achieved the highest average classification accuracy, and outperformed the best state-of-the-art baseline on each dataset by about 2%.

Motor Imagery Classification for Asynchronous EEG-Based Brain-Computer Interfaces

TL;DR

Addresses asynchronous MI-BCIs that must detect MI without triggers. Proposes SWPC, a two-module system with a prescreening stage and a multi-class MI classification stage, both trained with supervised learning followed by self-supervised learning for feature refinement; test data are ingested as sliding windows of length and gated by threshold . Across four BNCI-Horizon EEG datasets, SWPC achieves the highest average accuracy on both within-subject and cross-subject tasks and outperforms the best baselines by about 2 percentage points. This approach improves practicality of asynchronous BCIs and points to future work in transfer learning, test-time adaptation, and extending SWPC to other BCI paradigms.

Abstract

Motor imagery (MI) based brain-computer interfaces (BCIs) enable the direct control of external devices through the imagined movements of various body parts. Unlike previous systems that used fixed-length EEG trials for MI decoding, asynchronous BCIs aim to detect the user's MI without explicit triggers. They are challenging to implement, because the algorithm needs to first distinguish between resting-states and MI trials, and then classify the MI trials into the correct task, all without any triggers. This paper proposes a sliding window prescreening and classification (SWPC) approach for MI-based asynchronous BCIs, which consists of two modules: a prescreening module to screen MI trials out of the resting-state, and a classification module for MI classification. Both modules are trained with supervised learning followed by self-supervised learning, which refines the feature extractors. Within-subject and cross-subject asynchronous MI classifications on four different EEG datasets validated the effectiveness of SWPC, i.e., it always achieved the highest average classification accuracy, and outperformed the best state-of-the-art baseline on each dataset by about 2%.

Paper Structure

This paper contains 15 sections, 9 equations, 10 figures, 17 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Methodology
  3. Flowchart of SWPC
  4. Problem Setting
  5. The Prescreening Module
  6. The Classification Module
  7. Experiments
  8. Datasets
  9. Performance Evaluation
  10. Algorithms
  11. Results
  12. Ablation Study
  13. Parameter Sensitivity Analysis
  14. Offline Classification
  15. Conclusions and Future Research

Figures (10)

  • Figure 1: Flowchart of a closed-loop EEG-based BCI system.
  • Figure 2: Illustration of asynchronous MI classification. The user may switch between resting-state and MI at any unknown time.
  • Figure 3: SWPC for asynchronous MI-based BCIs.
  • Figure 4: Usage of the prescreening probability $\bar{p}_i$.
  • Figure 5: Training of the prescreening module.
  • ...and 5 more figures