BCI Performance Optimization Using EEG Band Power and Hybrid Deep Learning

Abstract: Brain-Computer Interface (BCI) systems rely on accurate interpretation of electroencephalogram (EEG) signals to understand human cognitive states. However, EEG signals are highly noisy, non-linear, and vary across individuals, which reduces classification accuracy. This paper proposes a real-time BCI performance optimization system using EEG band power features and a hybrid deep learning model. The system processes Alpha, Beta, and Gamma brainwave components to extract meaningful band power features. Additionally, EEG signals are converted into spectrogram images to capture time-frequency patterns. A hybrid Convolutional Neural Network (CNN) model is used to combine spectrogram-based image features with numerical band power features for improved classification. The system predicts cognitive states such as Relaxed, Normal, and High Stress. A real-time interactive dashboard is developed for visualization, prediction, and report generation. Experimental results demonstrate improved classification accuracy and robustness compared to traditional approaches. The proposed system enhances BCI reliability and can be applied in healthcare, mental monitoring, and neuro-adaptive systems.

Keywords: EEG, BCI, Deep Learning, CNN, Band Power, Spectrogram, Stress Detection, Neurotechnology1