Decoding multi-joint hand movements from brain signals by learning a synergy-based neural manifold.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Robotics

Background:

• Brain-computer interfaces (BCIs) show promise for restoring motor functions.
• Decoding complex natural movements like hand movements remains a significant challenge.
• Existing methods often decode individual joint movements, neglecting underlying motor synergies.

Purpose of the Study:

• To explore the role of motor synergies in complex hand movements.
• To develop a novel neural decoding framework for improved hand movement reconstruction.
• To investigate if decoding motor synergies offers advantages over traditional joint-level decoding.

Main Methods:

• Decomposition of complex hand movements into motor primitives (synergies).
• Learning a joint neural-motor representation of these synergies.
• Development and application of the SynergyNet framework for decoding spatiotemporal parameters.
• Comparison of SynergyNet against benchmark methods for hand movement decoding.

Main Results:

• Demonstrated that complex hand movements can be represented by motor synergies.
• Showcased that recruiting motor synergies via spatiotemporal parameters aids complex motor control.
• SynergyNet significantly outperformed existing benchmark methods in hand movement decoding accuracy.
• The proposed approach offers high interpretability in neural decoding.

Conclusions:

• Motor synergies are fundamental to complex hand movements and can be effectively decoded.
• Decoding spatiotemporal parameters of motor synergies offers a more effective approach than joint-level decoding.
• The SynergyNet framework provides a powerful and interpretable tool for brain-computer interface applications in motor control.