Single-Trial Detection With Magnetoencephalography During a Dual-Rapid Serial Visual Presentation Task

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Detecting brain responses to specific image classes is crucial for brain-machine interfaces (BMIs).
• Current rapid serial visual presentation (RSVP) systems offer gaze independence and high throughput, benefiting both healthy and disabled users.
• A challenge remains in improving the accuracy and efficiency of single-trial brain response detection.

Purpose of the Study:

• To develop and evaluate a novel dual-RSVP task paradigm for enhanced brain-machine interface performance.
• To assess the feasibility of single-trial detection of brain responses using magnetoencephalography (MEG) during the proposed task.
• To investigate the effectiveness of combining decisions from dual-stream presentations for improved target detection accuracy.

Main Methods:

• A dual-RSVP task was designed with two simultaneous image streams, one temporally delayed, assuming low target probability.
• Participants were instructed to identify images containing a person, shifting attention between streams based on target appearance.
• Magnetoencephalography (MEG) signals were recorded, and classification performance was evaluated using a Bayesian Linear Discriminant Analysis (BLDA) classifier after spatial filtering.

Main Results:

• Single-trial detection performance was assessed across individual streams and their combined decisions.
• Classification accuracy was compared using different channel sets (magnetometers, gradiometers).
• The study demonstrated high performance, with an area under the ROC curve exceeding 0.95 for single-trial detection.

Conclusions:

• The dual-RSVP task enables reliable single-trial target detection using MEG.
• Combining decisions from two trials significantly improves accuracy, approaching perfection for some subjects without extending experiment duration.
• This approach offers a promising advancement for brain-machine interface applications requiring precise visual stimulus detection.