Offline Lower-Limb Kinematic Decodification by Segments of EEG Signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Decoding human movement from neural signals is crucial for advanced prosthetics and assistive devices.
• Electroencephalography (EEG) offers a non-invasive method for capturing brain activity related to motor intentions.
• Understanding joint kinematics from EEG requires robust signal processing and decoding strategies.

Purpose of the Study:

• To investigate the efficacy of different electroencephalography (EEG) data processing strategies for decoding hip and knee joint angles.
• To compare the performance of multiple linear regression (MLR) models when applied to segmented versus unsegmented EEG data.
• To determine if task-specific segmentation enhances the accuracy of movement decoding from neural signals.

Main Methods:

• Low-frequency electroencephalography (EEG) components were recorded during the execution of hip and knee movement tasks.
• Three multiple linear regression (MLR) models were implemented to decode joint angles.
• EEG data were processed under two conditions: as a whole and divided into task-specific segments.

Main Results:

• Decoding of hip and knee angles from EEG signals was achieved.
• Segmentation of EEG data into task-specific segments resulted in improved decoding performance compared to using unsegmented data.
• The MLR models demonstrated varying performance based on the data processing condition.

Conclusions:

• Task-specific segmentation of EEG data significantly enhances the accuracy of decoding hip and knee joint angles.
• This finding has implications for developing more responsive and intuitive brain-computer interfaces for lower limb movement control.
• Optimizing EEG signal processing through segmentation is a key factor for successful neural decoding of motor actions.