Two Seconds to Speak: Increasing Communication Speed for fMRI-Based Brain-Computer Interfaces

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) offer motor-independent communication for individuals with lost motor function.
• Functional magnetic resonance imaging (fMRI)-based BCIs utilize hemodynamic brain activity from mental tasks.
• Improving the efficiency and speed of fMRI-BCIs is crucial for clinical applications due to the slow hemodynamic response.

Purpose of the Study:

• To enhance brain-computer interface (BCI) efficiency by discriminating brain activation patterns from short (2-second) mental tasks.
• To identify optimal mental task combinations for high-accuracy multi-class classification in fMRI-BCIs.
• To compare accuracy-based versus preference-based individualized task selection for yes/no communication paradigms.

Main Methods:

• Tested the reliability of discriminating distributed 3T-fMRI brain activation patterns from 2-second mental tasks across 2- to 7-class classifications.
• Identified optimal mental task combinations for maximizing classification accuracy.
• Evaluated individualized task selection strategies (accuracy-based vs. preference-based) for a yes/no communication task.

Main Results:

• Achieved 78% mean accuracy for 2-class decoding, with 3- to 7-class accuracies above chance.
• Mental calculation and spatial navigation tasks were most frequently linked to high decoding accuracy.
• Yes/no answers were encoded with high accuracy (83% accuracy-based, 81% preference-based) using individualized tasks.

Conclusions:

• Short mental tasks (2 seconds) can reliably evoke discriminable fMRI activation patterns, increasing BCI efficiency.
• Individualized task selection, whether accuracy- or preference-based, significantly enhances communication accuracy in fMRI-BCIs.
• This fMRI-BCI paradigm is suitable for diverse patient populations, offering potential for greatly improved clinical utility.