Denoising of EEG Signal based on Matlab Wavelet Toolbox and Estimation of Signal to Noise Ratio (SNR), Root Mean Square Error (RMSE) and Power Spectral Density (PSD)

Most signals in the real world contain noise. Any undesired signal that distorts the desired signal is considered noise in this context. Noise must be removed to obtain valuable information from the signal. Electroencephalography (EEG) is the primary signal of interest for the analysis of brain activity. EEG uses electrodes attached to the scalp to capture the electrical activity of the brain. From the diagnosis of neurological diseases to controlling external devices using brain-computer interface (BCI), EEG signal processing is of utmost importance. Nevertheless, artifacts created while recording the EEG signal in a real-world setting can significantly reduce the signal's effectiveness. The recorded signals are typically so severely noisecorrupted that they are useless and limit the use of the EEG signal. The wavelet transform (WT) is the most widely used and effective method for signal denoising with non-stationary signals, including electroencephalogram (EEG) and electrocardiogram (ECG). In this paper, a wavelet signal denoiser application is used, which is available in the wavelet toolbox of MATLAB. Signal denoising is implemented using different sets of combinations of parameters, such as mother wavelets, methods of thresholding, and rules of thresholding. The effectiveness of signal denoising is dependent on setting its control parameters in the best possible way, which is frequently done through experimentation. Fortunately, measuring parameters like the mean squared error (MSE), signal-to-noise ratio (SNR), and power spectral density (PSD) may be used to assess how well these factors work together. Later, we will compare the results of different combinations of control parameters and determine the best one using the values of the measuring parameters. The real-time dataset is taken from the Mind wave Mobile 2 headset, which is an effective and portable EEG device. It uses Bluetooth connectivity to transfer the EEG signal from the device into a CSV file on our mobile phone. This noisy EEG signal is decomposed into up to four levels. This EEG signal decomposition results in approximate and detailed coefficients, which can be further used for the process of feature extraction from the EEG signal.