| Abstrakt: |
Electroencephalography (EEG) is a non-intrusive method used to capture electrical potential generated by brain neurons, which is crucial for diagnosing neurological disorders like epilepsy, sleep disorders, brain tumours, and dementia. However, EEG signals can sometimes be contaminated with undesired signals, known as artifacts. These artifacts are generated by retinal dipole movement and eyelid motion, resulting in spikes with significant amplitudes and distortions in EEG recordings. Researchers have explored wavelet-based techniques to remove EOG artifacts from EEG signals, with Haar, Symlet, and Daubechies being common wavelets. However, selecting the most suitable wavelet remains a challenge. The proposed study found that db7 is the optimal wavelet for removing EOG artifacts, with the lowest RMSE value of 54.97 and a calculation time of 0.5798 s. The proposed wavelet-independent component analysis approach outperforms the existing independent component analysis, specifically Fast-ICA, with a promising computation time (ΔCt) of 6.27 s. The proposed approach aims to enhance the quality of EOG signals by effectively reducing noise through the selection of an appropriate mother wavelet function. |
Full text from SpringerLink