A Non-Invasive, MRCP-Based BCI for Online Communication

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Individuals with severe motor impairments often lack reliable communication methods.
• Brain-computer interfaces (BCIs) offer a promising avenue for restoring communication abilities.
• Existing BCIs may require extensive calibration or external cues, limiting intuitive use.

Purpose of the Study:

• To develop and evaluate a novel BCI system for asynchronous detection of movement intent from EEG signals.
• To create a task-agnostic framework for controlling spelling applications, emulating a button press.
• To assess the system's performance in enabling communication for individuals with motor impairments.

Main Methods:

• A BCI system was developed to detect movement-related cortical potentials (MRCPs) from EEG data.
• The system asynchronously detects self-initiated movements without external cues.
• A spelling interface using row-column scanners (3x3 and 5x5) was controlled by the BCI for typing five-letter words.

Main Results:

• The MRCP-based classifier demonstrated consistent detection performance across different interface configurations.
• Participants achieved an average true positive rate (TPR) of 54.4% and correctly selected 60.9% of target characters.
• A five-letter word was successfully spelled in 41.7% of attempts, with low false positive rates.

Conclusions:

• The developed BCI system shows potential as a non-invasive communication aid for individuals with motor impairments.
• The task-agnostic, MRCP-based approach offers robust and adaptable control for various applications.
• This research lays the groundwork for future home-use BCIs with intuitive, voluntary control and minimal calibration.