RT-NET

RT-NET reconstructs neural activity in real time from high-density electroencephalography (hdEEG) to enable online source-projected analysis for source-based neurofeedback and brain-computer interface (BCI) applications.

Key Features:

• Spatial filter calibration: Estimates a spatial filter for artifact removal and source activity reconstruction using a calibration dataset.
• Real-time filter application: Applies the calibrated spatial filter to incoming hdEEG data with minimal latency and reduced computational load.
• Data acquisition and streaming: Uses the Lab Streaming Layer to acquire raw EEG data from multiple amplifiers and to stream processed outputs to external applications.
• Visualization: Provides online visualizations of neural activity in 2D and 3D, including projections onto a 3D cortical model.
• Performance parity: Produces neural activity estimates comparable to traditional offline methods.
• Targeted-area analysis: Supports online experiments focused on one or two specific brain areas.

Scientific Applications:

• Source-based neurofeedback: Enables closed-loop neurofeedback experiments based on source-projected hdEEG activity.
• Brain-computer interfaces (BCI): Supports real-time source-level signals for BCI paradigms.
• Neural dynamics research: Facilitates large-scale investigations of neural dynamics in healthy and diseased human brains using online hdEEG source analysis.

Methodology:

Estimate a spatial filter for artifact removal and source activity reconstruction from a calibration dataset; apply the spatial filter in real time to incoming hdEEG; acquire raw EEG from multiple amplifiers and stream processed outputs via the Lab Streaming Layer.