Muscle artifact removal in ictal scalp-EEG based on blind source separation

Ahmad Karfoul (1) , Amar Kachenoura (2) , Laurent Albera (2, 3) , Doha Safieddine (2) , Anca Pasnicu (4) , Fabrice Wendling (2) , Lotfi Senhadji (2) , Isabelle Merlet (2)

1 Faculty of Mechanical and Electrical Engineering [Homs]
2 LTSI - Laboratoire Traitement du Signal et de l'Image
3 PANAMA - Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio
4 Service de Neurologie [Rennes] = Neurology [Rennes]

Electroencephalogram (EEG) recordings are often contaminated with muscle artifacts. These artifacts obscure the EEG and complicate its interpretation or even make the interpretation unfeasible. This paper focuses on the particular context of extraction of low-voltage rapid ictal discharges from ictal scalp-EEG activity cantaminated by muscle artifact. In this context our aim was to evaluate the ability of Independent Component Analysis (ICA) and Canonical Correlation Analysis (CCA), to remove muscle artefacts from surface EEG sig- nals. The efficiency of ICA and CCA to correct the muscular artifact was evaluated both on simulated data and on real data recorded in an epileptic patient. The obtained results show that some ICA methods and CCA removed successfully the muscle artifact without altering the recorded underlying ictal activity.

Domaines: Ingénierie biomédicale; Neurosciences [q-bio.NC]

Ahmad Karfoul, Amar Kachenoura, Laurent Albera, Doha Safieddine, Anca Pasnicu, et al.. Muscle artifact removal in ictal scalp-EEG based on blind source separation. 6th European Conference of the International Federation for Medical and Biological Engineering, Sep 2014, Dubrovnik, Croatia. 4 p. ⟨hal-01015798⟩