EEG based dynamic RDS recognition with frequency domain selection and bispectrum feature optimization.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Stereopsis is crucial for daily life and is studied using random-dot stereograms (RDS).
• Electroencephalogram (EEG) is a noninvasive technique for visual cognition research.
• Dynamic RDS (DRDS) presents a challenge for EEG signal classification.

Purpose of the Study:

• To propose a new methodology for classifying EEG signals evoked by DRDS.
• To enhance the accuracy of DRDS recognition using EEG data.
• To validate the effectiveness of the proposed method.

Main Methods:

• Wavelet packet transform (WPT) for signal decomposition into frequency bands.
• Bispectral eigenvalues of differential signals (BED) for feature extraction.
• Selection of optimal sub-bands and reconstruction for improved classification.

Main Results:

• The WPT-BED method achieved a highest classification accuracy of 84.38% and an average accuracy of 73.98%.
• Analysis focused on active channels within the visual cortex.
• The method demonstrated superior performance compared to existing techniques.

Conclusions:

• The WPT-BED methodology effectively distinguishes EEG signals evoked by DRDS.
• This research confirms the feasibility of recognizing DRDS using EEG.
• The findings contribute to advancements in visual cognition and brain-computer interfaces.