Epileptic EEG Signal Denoising Enhancement Using Improved Threshold Based Wavelet Method
P. Divya, B. Aruna Devi · 2021 International Conference on System, Computation, Automation and Networking (ICSCAN) · 2021
Epilepsy is a disease of the brain characterized by enduring predisposition to generate epileptic seizures. It is one of the most common neurological illness affecting individuals of any age. The diagnosis of epilepsy is made primarily on clinical grounds supporting investigation, which includes Electroencephalography (EEG), Electrocardiography (ECG) and Electrocorticography (ECoG), allowing both lightweight analysis and extensive analysis to provide more accurate and reliable decisions. In this work, two algorithms for EEG signal denoising is proposed. One method uses Savitzky-Golay (S-G) smoothing filter to remove noisy data from EEG signal in time domain and second method uses improved wavelet thresholding function to discard redundant data in wavelet domain. Raw EEG signals are decomposed into different sub bands by using wavelet transform. Only detailed sub bands are filtered by using thresholding method. Performance of the proposed denoising methods are compared by computing statistical value. Results showed that the improved wavelet thresholding method provides better results when compared to S-G smoothing filter. A new algorithm for denoising based on wavelet thresholding function is proposed.