A new method for removing Electrical Vestibular Stimulation induced artifacts from Electro-Encephalographic recordings

• Main Author: Adib, Mani
• Language: English
• Published: University of British Columbia 2012
• Online Access: http://hdl.handle.net/2429/43026

In this work, we present a new method for removing artifacts from scalp Electro-Encephalography (EEG) signals recorded during Electrical Vestibular Stimulation (EVS). Using EVS, we stimulate the vestibular nerves, which can affect different regions in the brain via the interconnection of the vestibular system with some regions in the brain. As a result, some of the brain functions can be altered during the EVS application. Throughout its long history, EVS has been found as an interesting research tool in physiology and neurology. Various applications of EVS have been implemented in the health-care industry and also in other industries such as entertainment. Although there have been many advances in the EVS applications, it still remains a challenging problem to understand how the EVS stimulus acts as an input to the brain and how the brain responds. In this study, we monitored and recorded the brain activities during the application of EVS, using EEG. The recorded EEG data during EVS application, contain the information that elicit the EVS induced responses. However, the distribution of the EVS current throughout the scalp generates an artifact on the EEG signals. To analyze the EEG and study the brain functions during EVS, we have to eliminate this artifact. We developed a method to remove this artifact by estimating the contribution of the EVS current in the EEG signals at each electrode. The proposed method is a hybrid method, which combines time series regression and wavelet decomposition methods to estimate the artifact and remove it. Wavelet transform was employed to project the recorded EEG signal into various frequency bands and then the regression method was used to estimate the EVS current distribution in each frequency band separately. We optimized the proposed method using simulated data. Then we assessed the performance of our method and compared it to the other well accepted artifact removal methods, using both simulated and real data. The results show that the proposed method has better performance compared to the others, in terms of achieving higher signal to artifact ratio and introducing less distortion to the original EEG signals.