Detection and removal of ocular artifacts using Independent Component Analysis and wavelets

Abstract

In this paper a novel approach for ocular artifact (OA) removal is proposed in which a combination of Independent Component Analysis and wavelet-based noise reduction is utilized for detection and removal of OA. At the first stage, independent basis functions attributed to OA are computed using FastICA algorithm. This is followed by designing a wavelet basis function which is tuned to have sufficient similarity in its waveform to the independent basis functions of OA. We then utilize the designed wavelet for signal decomposition in a standard discrete wavelet transform where by deleting the approximation and summing up the details of signal decomposition, we arrive at a sufficiently artifact-free EEG signal. The approach excludes thresholding challenges of wavelets and works both for eye blinks and eye movements. Applying our algorithm to 420 4-s EEG epochs, the method exhibits high performance for the removal of OA artifacts. Our wavelet design method for noise reduction can be extended to the removal other types of EEG artifacts Keywords-discrete wavelet transform (DWT), independent component analysis (ICA), wavelet design, electroencephalogram (EEG), ocular artifact (OA)

A deep descriptor for cross-tasking EEG-based recognition

“…Regardless of the application, preprocessing is required since the signal is very susceptible to noise. Several approaches have been used in the literature for data processing, such as Common Spatial Pattern (CSP) ( Yong, Ward & Birch, 2008 ), Wavelet transforms ( Kumar et al, 2009 ; Ghandeharion & Ahmadi-Noubari, 2009 ), and Independent Component Analysis (ICA) ( Delorme, Sejnowski & Makeig, 2007 ).…”

Real-Time Low-Latency Estimation of the Blinking and EOG Signals

“…Another method of the event detection (a saccade, a blink) is the wavelets transform [Ghandeharion & Ahmadi-Noubari (2009);Reddy et al (2010)]. The saccade and blink are well defined in time domain and they have a limited length in time.…”

Automatic Extraction System for Common Artifacts in EEG Signals Based on Evolutionary Stone’s BSS Algorithm

“…BSS and parallel factor analysis (PFA) are integrated to reject the EEG artifacts [8,9]. Wavelet transforms (WT) with independent component analysis (ICA) and statistical autoregressive moving average model have been used to reject the artifacts [10]. Pesin [11] demonstrates a novel approach to recognize and reject eye blink artifacts from EEG system based on an integration between wavelet technique and FastICA to expose the temporal position of eye blink and then remove it.…”