研究目的
To propose and evaluate a self-flickering visual stimulus based on visual illusion (windmill pattern) for SSVEP-based BCI systems to reduce eye fatigue and increase the number of commands.
研究成果
The self-flickering visual stimulus based on windmill patterns effectively induces SSVEP responses, allowing for three BCI commands with an average classification accuracy of 80.5%. It reduces eye fatigue and can increase command options for existing SSVEP-based BCIs, making it suitable for practical applications. Future work should explore additional windmill patterns, bilateral response utilization, and hybrid approaches to enhance performance.
研究不足
The accuracy of the proposed system was lower than general SSVEP-based BCI systems in non-fatigue states. Limitations include the small sample size (five subjects), potential for optimization in feature extraction and classification algorithms, and the need for further validation with more subjects and in varied conditions.
1：Experimental Design and Method Selection:
The study employed a visual illusion phenomenon using windmill patterns to induce SSVEP responses. SSVEP detection algorithms were implemented in LabVIEW for real-time BCI command classification.
2：Sample Selection and Data Sources:
Five healthy volunteer subjects (mean age 23 ±
3：3 years for brain response observation and 23 ± 8 years for testing) without prior VEP-BCI experience were used. EEG signals were recorded from occipital regions. List of Experimental Equipment and Materials:
Brainmaster Discovery 24E system with 19-channel EEG cap, BIOPAC system EEG amplifiers, NI USB 6009 data acquisition card, 14-inch LCD screen for visual stimulus display, and NeuroGuide software for brain mapping.
4：Experimental Procedures and Operational Workflow:
Subjects stared at the center of windmill patterns (24, 48, and 96-windmill) on an LCD screen. EEG signals were recorded from O1 and O2 electrodes with a sampling rate of 256 Hz, filtered using band-pass (1-35 Hz) and notch (50 Hz) filters. Calibration involved baseline PSD measurements, feature extraction during stimulus, and decision making for command classification over 4-second trials.
5：Data Analysis Methods:
Power spectral density (PSD) analysis at harmonic frequencies (8, 16, 24, 32 Hz) was performed. Features were extracted and classified using a simple decision rule based on summation scores, with accuracy calculated from 60 trials per subject.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
