Decoding fingertip trajectory from electrocorticographic signals in humans.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Brain-Computer Interfaces

Background:

• Brain-machine interface (BMI) technology is advancing neuroprosthetics, with successful trajectory prediction for elbow and wrist movements.
• Predicting hand trajectory and classifying gestures is a growing area of interest in BMI research.
• Precise finger motion trajectory prediction remains a significant challenge in the field.

Purpose of the Study:

• To propose and evaluate a novel method for predicting fingertip motion trajectories.
• To decode fingertip movements directly from electrocorticographic (ECoG) signals recorded from the human cortex.
• To assess the generalizability and accuracy of the proposed prediction method.

Main Methods:

• Developed a method to predict fingertip trajectories from electrocorticographic (ECoG) signals.
• Recorded ECoG data while a patient performed three-finger extension/flexion tasks.
• Validated the method's generalizability using open datasets from BCI Competition IV.

Main Results:

• Achieved high prediction accuracy for fingertip trajectories, with average Pearson's correlation coefficients ranging from 0.83 to 0.90.
• Obtained normalized root-mean-square errors between 0.24 and 0.48, indicating precise decoding of actual trajectories.
• Demonstrated comparable prediction accuracy to existing methods on benchmark datasets, confirming method's generalizability.

Conclusions:

• The proposed ECoG-based method effectively predicts fingertip motion trajectories.
• This approach holds significant potential for enhancing the precision of neuroprosthetics and robotic control systems.
• The demonstrated generalizability suggests broad applicability across different users and datasets.