An Insightful Analysis of Preprocessing Methods Used in EEG Signals for Computer-Assisted Cognitive Domain

The most popular non-intrusive approach for analyzing electrical brain activity is electroencephalography (EEG), which is employed comprehensively in mental neuroscience and medical diagnostics. EEG readings are frequently contaminated with various types of noise and artifacts, which can conceal the relevant brain activity. It has been proven that preprocessing strategies are necessary to improve data quality, reduce noise or artifacts, and prepare the data for proper analysis and interpretation. Therefore, substantial research over the last 20 years has concentrated on discovering methods for dealing with these artifacts during the preprocessing step. Nevertheless, it remains a focus of present research because no particular recognized artifact removal method is comprehensive or ubiquitous. The paper describes a qualitative study of current state-of-the-art of substantial preprocessing pipeline stages for EEG signal including filtering, sampling rate conversion, artifacts removal, channel elimination, and re-referencing. Subsequently, the finite impulse response (FIR) filter with rectangular, hamming, and Blackman windows are compared based on their average side lobe peak in frequency domain and black window outperformed among the all. It is considerable that this study provides the researchers with a contemporary perspective on preprocessing pipeline of EEG signals and acts as a basic guideline in general.