Stable online control of an electrocorticographic brain-computer interface using a static decoder.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Brain-computer interfaces (BCIs) enable direct communication between the brain and external devices.
• Electrocorticography (ECoG) records brain activity from the surface of the brain, offering high spatial and temporal resolution.
• Understanding neural control strategies is crucial for advancing BCI technology.

Purpose of the Study:

• To investigate the learning curve and co-adaptive training of a Rhesus macaque using a BCI.
• To assess the stability and duration of BCI control.
• To explore the neural mechanisms and strategies underlying BCI operation.

Main Methods:

• Implemented a BCI system using electrocorticographic (ECoG) signals from the motor cortex of a Rhesus macaque.
• Utilized an optimal linear estimation (OLE) method for decoding neural signals into two-dimensional cursor movements.
• Employed a co-adaptive training scheme to facilitate learning and control acquisition.

Main Results:

• The macaque achieved accurate and sustained BCI control within 4-5 days of co-adaptive training.
• Stable control performance was maintained for 28 days with constant decode parameters.
• Analysis of neural strategies revealed insights into feature clustering and coordination for kinematic control.

Conclusions:

• BCI systems can be effectively learned and controlled by non-human primates through co-adaptive training.
• Neural decoding parameters demonstrate stability over extended periods.
• Investigating neural strategies provides a deeper understanding of brain function during BCI use.