Utilizing movement synergies to improve decoding performance for a brain machine interface.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Robotics

Background:

• High-degree of freedom (DOF) brain-machine interfaces (BMIs) face challenges in providing users with independent control of complex prostheses.
• Movement synergies can replicate complex hand postures with fewer components, suggesting a reduced dimensionality for control.

Purpose of the Study:

• To investigate the effectiveness of principal component analysis (PCA) in reducing the dimensionality of high-DOF arm and hand movements for BMI control.
• To determine if decoding neural activity in a lower-DOF eigen-reach space improves BMI performance.

Main Methods:

• Applied PCA to decompose the high-DOF joint space of the arm and hand into a lower-DOF eigen-reach space.
• Analyzed neural activity decoding within this reduced eigen-reach space.
• Compared decoding performance with and without pre-processing using PCA.

Main Results:

• PCA effectively captures most of the movement variance in a lower-dimensional eigen-reach space.
• Decoding neural signals in the eigen-reach space requires controlling fewer variables.
• Preliminary findings suggest potential improvements in decoding performance by using PCA before neural decoding.

Conclusions:

• Dimensionality reduction via PCA is a promising strategy for enhancing BMI control of high-DOF prostheses.
• Simplifying the control space through eigen-reach allows for more intuitive and potentially more effective prosthetic limb operation.
• Further research is warranted to fully validate the benefits of PCA in real-time BMI applications.