Total Variation Denoising techniques for artifact removal from EEG signals

摘要

In detecting brain activity and behavior, the Electroencephalogram (EEG) plays a key role. The EEG signals recorded are almost always corrupted by artifacts and hence affect the EEG signal analysis. Therefore, it is highly essential to devise techniques to extract noise-free EEG data from the recorded EEG signals. The performance efficiency of two variation denoising methods: Simultaneous Low-Pass Filtering/Total Variation Denoising(LPF/TVD) and Transient Artifact Reduction Algorithm (TARA) in artifact removal from Schizophrenia (SZ) patient’s and healthy adolescent’s EEG signals are being evaluated. The artifact removal technique of TARA is a modified genre of Simultaneous LowPass Filtering and Total Variation Denoising (LPF/TVD). It has been observed that TARA performs well in denoising the signals from a group of 45 SZ and 39 healthy control EEG signals. The efficiency of the methods is evaluated using the index of Signal to noise ratio (SNR). A high SNR value for TARA shows the efficiency of this method in removing the artifacts from SZ EEG signals.