Abstract

Although electroencephalography, or EEG, is a method for monitoring brain activity, it is easily able to detect undesired noise, such as electrical interference, muscle contractions, and even eye blinks. Accurately interpreting the brain signals can be challenging due to these distractions. To clean up EEG readings, researchers have developed a number of techniques, such as spectral subtraction, wavelet transforms, and deep learning techniques like autoencoders. PyZaplinePlus, meegkit, and noise reduce are some of the useful Python tools that have recently surfaced and provide automated methods for removing these undesirable sounds. For improved results, some sophisticated models, such as IC-U-Net, combine neural networks and statistical techniques. But these techniques frequently call for intricate configurations and a large amount of processing power. In our study, we give a more direct and effective point of view using two types of filters: a band-pass filter and a notch filter. The band-pass filter keeps the frequency range crucial for brain activity, which is between 0.5 and 40 Hz. Meanwhile, the notch filter eliminates the 50 Hz electrical noise from power lines. After testing our method, we saw a major improvement in signal quality. The signal-to-noise ratio, which points how clear the signal is, changed from about 7.5 dB to over 11 dB in several trials. This shows that even simple methods can enhance signal quality without expensive or complicated tools. Our approach is user-friendly for beginners working with EEG signals, making it useful for both learning and research. Looking forward, we aim to enhance this method further by incorporating improved filters or machine learning techniques to reduce noise even more.