Detection of reaching intention using EEG signals and nonlinear dynamic system identification

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Low-frequency electroencephalography (EEG) signals are linked to movement preparation.
• These signals offer potential for brain-machine interface (BMI) applications.

Purpose of the Study:

• To detect movement intention from EEG signals prior to self-paced arm reaching movements.
• To develop a novel decoding method for BMI applications.

Main Methods:

• A nonlinear dynamic multiple-input/single-output (MISO) model using Volterra structures was developed.
• Input selection from EEG channels and threshold optimization were performed.
• Model performance was assessed using receiver operating characteristic curves and Mann-Whitney statistics.

Main Results:

• Movement intention was detected approximately 500 ms before movement onset.
• The model achieved high accuracy (96.37% ± 0.94%), sensitivity (77.93% ± 4.40%), and specificity (98.52% ± 1.19%).

Conclusions:

• The developed model effectively decodes movement intention from EEG signals.
• The model's output can be translated into control signals for neuroprosthetics and exoskeletons.