Decoding of Self-paced Lower-Limb Movement Intention: A Case Study on the Influence Factors

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Brain-machine interfaces (BMIs) are emerging as rehabilitation tools for motor-impaired individuals.
• Cerebral activity engagement enhances neuroplasticity and aids mobility restoration.
• Research has primarily focused on upper-limb movement intention, with less attention on lower-limb decoding.

Purpose of the Study:

• To develop and evaluate a BMI system for decoding self-paced lower-limb movement intention.
• To compare different processing methods (MRCP vs. SMR) and frequency bands for decoding accuracy.
• To analyze feature discriminant power and brain modulations for optimizing BMI design.

Main Methods:

• Ten healthy subjects participated in experiments decoding lower-limb movement intentions.
• Movement types (dorsiflexion, plantar flexion), limb side (left, right), processing methods (MRCP, SMR), and frequency bands were varied.
• Single-trial and sample-based performance metrics, including Area Under the Curve (AUC), were estimated.

Main Results:

• The MRCP-based method achieved an average AUC of 91.0 ± 3.5%, significantly outperforming the SMR-based method (68.2 ± 4.6%).
• The best performance was observed for left plantar flexion using time-series analysis on the MRCP band.
• MRCP-based decoding showed cross-subject consistency in discriminant power, unlike the SMR-based method.

Conclusions:

• The developed BMI effectively decodes lower-limb movement intention, particularly using MRCPs.
• The MRCP-based approach offers reliable and consistent decoding for potential rehabilitation applications.
• Findings provide insights for designing effective brain switches to control external robotic devices in rehabilitation settings.