Research And Application Of The EEG Recording Preprocessing

Electroencephalogram (EEG) is the spontaneous electrical activity recording of brain cells from scalp. It is important to remove artifacts from EEG recording for further study. Independent Component Analysis (ICA) is a kind of adaptive blind sources separation methods, which at present becomes the focus of signal-processing and signal-analysis field in the world. And adaptive filter technique has the power self-study and strong self-track abilities. These characteristics make it can detect and process the stationary or non-stationary signals. This paper focus on some key topics described as following:(1) Automatically isolate and remove the artifacts from EEG recording in ICA Based on the ICA of several signals, an automatic approach is presented toisolate and remove the electrocardiogram, respiration waveform artifacts and electrocardiogram (ECG) from the EEG recordings. Power spectrum as the criterion, to compute the proportion of the artifacts frequency band energy to the total energy for every component, the artifact component can be separated with the value of proportion. Through the experiment it is validated that this method can separate the artifacts component effectively. FastICA is adapted in the experiment. The results of restructure show that ICA adapts the signals with lots of electrode.(2) Non-linear ICA: MISEPAs a kind of complicated biology signal, analyzing EEG in the linear field is not enough. MISEP is a method for linear and nonlinear ICA. It is an extension of the well-known Infomax method, in two directions: (a) handling of nonlinear mixtures, and (b) learning the nonlinearities to be used at the outputs. The method can therefore separate linear and nonlinear mixtures of components with a wide ange of statistical distributions. Through the test we find that the non-function should be changed, and the stability is not good.(3) Several references adaptive noise canceller (ANC)Based the single ANC several references'ANC is presented in this paper. With the ECG artifacts and respiration artifacts in the EEG recording, a double references ANC is designed. In this method RLS algorithm is adopted to adjust the weights of filter for its good astringency. The result of experiment shows that several ANC can filter the several artifacts and hold the useful signal.