SO SÁNH HIỆU SUẤT GIỮA BỘ LỌC FIR VÀ LMS TRONG XỬ LÝ NHIỄU TÍN HIỆU ĐIỆN NÃO ĐỒ EEG

Abstract

This paper compares the performance of two electroencephalogram (EEG) signal filtering methods: the Finite Impulse Response (FIR) low-pass filter and the Least Mean Squares (LMS) adaptive filter, in terms of their ability to remove 50 Hz noise while preserving key frequency bands such as Delta, Theta, Alpha, and Beta. The evaluation metrics include the signal-to-noise ratio (SNR), mean squared error (MSE), processing time, harmonic distortion, and phase delay. The results indicate that while both methods significantly reduce noise, the FIR filter outperforms the LMS filter in terms of noise removal and preservation of EEG signals, demonstrating good accuracy in both time and frequency domains. The FIR filtering method proves superior in maintaining EEG signal integrity, while the LMS technique retains advantages in dynamic noisy environments, effectively reducing noise in the low and mid-frequency ranges. This study provides important insights into selecting the appropriate filtering method to enhance signal quality, minimize noise, and improve reliability in EEG analyses.