Decoding cortical responses from visual input using an endovascular brain-computer interface.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Medical Devices

Background:

• Implantable neural interfaces are crucial for understanding brain function.
• High-quality brain signal recording is essential for advancing neuroscience.
• Minimally invasive techniques are sought for neural recording.

Purpose of the Study:

• To evaluate the feasibility of an endovascular neural interface (ENI) for recording cortical activity.
• To assess the interpretability of visual cortex signals captured by an ENI.
• To compare ENI performance with traditional subdural electrode grids.

Main Methods:

• A sheep model (n=5) was used for recordings.
• Visually evoked potentials were recorded using an ENI and an electrocorticography (ECoG) grid.
• Machine learning models decoded visual stimuli categories (color, contrast, direction, frequency).

Main Results:

• The ENI recorded interpretable cortical activity from the visual cortex.
• Decoding performance was lower than ECoG but significantly above chance for several visual features.
• This marks the first report of visually evoked neural activity using a minimally invasive ENI.

Conclusions:

• Implantable macro-electrodes, including ENIs, provide sufficient signal definition for discerning visual percepts.
• Endovascular and intracranial placements are viable for capturing neural signals.
• ENIs represent a promising minimally invasive option for neural recording.