Blind source separation, wavelet denoising and discriminant analysis for EEG artefacts and noise cancelling

This paper proposes an automatic method for artefact removal and noise elimination from scalp electroencephalogram recordings (EEG). The method is based on blind source separation (BSS) and supervised classification and proposes a combination of classical and news features and classes to improve artefact elimination (ocular, high frequency muscle and ECG artefacts). The role of a supplementary step of wavelet denoising (WD) is explored and the interactions between BSS, denoising and classification are analyzed. The results are validated on simulated signals by quantitative evaluation criteria and on real EEG by medical expertise. The proposed methodology successfully rejected a good percentage of artefacts and noise, while preserving almost all the cerebral activity. The “denoised artefact-free” EEG presents a very good improvement compared with recorded raw EEG: 96% of the EEGs are easier to interpret.

• Extended evaluation of several BSS algorithms on physiologically plausible EEGs.
• Original feature set extraction for source classification, containing physiologically significant features obtained from the estimated mixing matrix and from time and frequency characteristics of the sources, without using EOG electrodes.
• Wavelet denoising introduction and analysis of its interactions with the others elements of the processing chain (BSS and classification).
• Processing chain validation on a large database of real EEGs, using objective criteria (classification rate) and a systematic medical evaluation.