The Analysis And Ocular Artifacts Removal From EEG Representing Mental Arithmetic Difficulty Levels

Mental Arithmetic(MA)tasks is one of the common cognitive tasks.According to the complexity of MA tasks and the cognitive ability of task performers,there have three different cognitive strategies probably.For complex mental arithmetic problems with no direct answers in the brain,it is necessary to require Working Memory(WM)of task performers.Task performers'mental arithmetic ability,the arithmetic complexity,and the time limit for answering arithmetic question would determine their mental workload level together,which may cause different degrees of acute psychological stress.Therefore,to research their brain state when task performers are performing MA tasks of two levels of complexity would reveal the mental arithmetic strategy and the psychological and physiological mechanisms which closely related to psychological stress and working memory of MA tasks,which are important in the field of cognitive neuroscience.EEG is a method of detecting cortical activity which widely used in the field of cognitive research.Its millisecond-level time resolution help researchers to effectively reveal the corresponding cognitive processes of the brain by analyzing their power distribution properties and dynamic time course of brain oscillations in specific frequency bands.However,it is controversial about power changes of theta rhythm,alpha rhythm and beta rhythm in MA tasks of different difficulty levels.Secondly,the dynamic properties of theta,alpha and beta waves in different difficulty mental arithmetic tasks are still not known yet.In addition,in order to obtain the reliable cognitive research results based on single-channel EEG,it is necessary to propose an algorithm which used for removing ocular artifacts from single-channel EEG.In this paper,multi-channel EEGs were recorded in 19 healthy subjects performing MA task.Five subjects drop out of the experiment,14 subjects finish the experiment and their quality EEG data were available for subsequent analysis.In this paper,the power spectral estimation and time-frequency analysis were employed on preprocessed EEG theta rhythm,alpha rhythm and beta rhythm of MA tasks corresponding to the two different arithmetic complexity.In order to remove ocular artifacts in single-channel EEG data,an algorithm which based on stationary wavelet tranform and local filtering for eliminating ocular artifacts in EEG signals is proposed.The details are as follows.1.Research on the power distribution properties of EEG theta,alpha and beta rhythm in MA tasks of two difficulty levels.To filter and remove ocular artifacts of 10 channels EEG data from the prefrontal cortex and parietal lobe which processed by ICA algorithm firstly.Secondly,the Welch power spectral estimation algorithm and baseline normalization were employed on preprocessed EEG data for gaining the relative power spectra density(rPSD)of MA tasks corresponding to the two different arithmetic complexity.Finally,using 1 Hz as the unit of bandwidth,paired T-tests were performed on rPSD of MA tasks of two levels of complexity which cover a frequency range of 4-30 Hz for investigating the frequency range which can represent the significant differences in MA tasks of two levels of complexity.The results show that there are significant differences in EEG prefrontal and parietal rPSD of MA tasks of two levels of complexity within the 10-25 Hz and 9-25 Hz frequency ranges respectively,which include upper alpha and lower beta bands.2.Research on the dynamic time course of EEG theta,alpha and beta rhythm in MA tasks.Using the Morse wavelet,continuous wavelet transform is performed on the EEG signal of the first 5 seconds of each simple MA tasks or difficult MA tasks to obtain the wavelet coefficient energy corresponding to the simple MA tasks or difficult MA tasks.The results show that there is no significant difference between the power of the above three EEG rhythm in the 0.5-1 second period of the MA tasks corresponding to the two different arithmetic complexity,and higher power of alpha and beta rhythm in the last 3.5 seconds period of the simple MA tasks.The results are consistent with those reported in the literature about alpha inhibition in high intensity cognitive tasks.3.The proposed OD-SWT algorithm combines the stationary wavelet transform with the use of statistical thresholds for artifact segment detection to reject ocular artifacts from single-channel EEG signals.The results show that its effectiveness on ocular artifacts removal can be compared with ICA when OD-SWT algorithm combines with the optimal statistical threshold.OD-SWT algorithm is a practical techniques which are suitable for removing ocular artifacts from single-channel EEG signals.The rPSD changes in the alpha and beta bands are expected to be used as neurophysiological indicators for characterizing MA task complexity.The proposed OD-SWT algorithm provides a new idea for ocular artifacts removal from single-channel EEG signals.