Litcius/Paper detail

Ocular Artifacts Elimination From Multivariate EEG Signal Using Frequency-Spatial Filtering

Abhijit Bhattacharyya, Aarushi Verma, Radu Ranta, Ram Bilas Pachori

2022IEEE Transactions on Cognitive and Developmental Systems14 citationsDOIOpen Access PDF

Abstract

The electroencephalogram (EEG) signals record electrical activities generated by the brain cells and are used as a state-of-the-art diagnosis tool for various neural disorders. However, the unwanted artifacts often contaminate the recorded EEG signals and disturb the interpretation of the neuronal activity. This article aims to propose an efficient automatic method to eliminate the ocular artifacts (OAs) from the multichannel EEG signals with novel frequency-spatial filtering. The method combines dictionary-based spatial filtering and frequency-based signal decomposition method, namely, empirical wavelet transform (EWT). The artifact dictionary needed for spatial filtering is isolated from the raw data by: 1) selecting the contaminated channels and 2) frequency-domain filtering. More precisely, the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta $ </tex-math></inline-formula> -rhythms of identified highly contaminated channels are selected and placed into an artifact dictionary. Afterward, the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta $ </tex-math></inline-formula> -rhythms of multichannel EEG signals are spatially filtered using the built dictionary to seclude the OAs within a limited number of components. Furthermore, the artifact components are eliminated and clean <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta $ </tex-math></inline-formula> -rhythms are recovered using the inverse spatial filtering technique. Finally, the clean <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta $ </tex-math></inline-formula> -rhythms are combined with other EEG rhythms to reconstruct the OA-free signals. The proposed method is applied to OA-contaminated synthetic and real multichannel EEG signals with a convincing performance as compared to state-of-the-art approaches. The proposed method removes the OAs without affecting the background EEG information. The proposed method can ease sensor signal interpretation and further processing, e.g., for BCI applications.

Topics & Concepts

Artifact (error)NotationPattern recognition (psychology)Computer scienceSIGNAL (programming language)Spatial filterArtificial intelligenceElectroencephalographySignal processingSpeech recognitionMathematicsDigital signal processingArithmeticPsychologyPsychiatryComputer hardwareProgramming languageEEG and Brain-Computer InterfacesBlind Source Separation TechniquesNeural dynamics and brain function
Ocular Artifacts Elimination From Multivariate EEG Signal Using Frequency-Spatial Filtering | Litcius