RT info:eu-repo/semantics/article
T1 Calibration-free Ocular artifact reduction in EEG signals using a montage-independent deep learning model
A1 Marcos Martínez, Diego
A1 Pérez Velasco, Sergio
A1 Martínez Cagigal, Víctor
A1 SantaMaría Vazquez, Eduardo
A1 Hornero Sánchez, Roberto
K1 Deep learning
K1 Ocular artifacts
K1 Electroencephalography
K1 Brain–computer interfaces
K1 Electrooculography
K1 32 Ciencias Médicas
K1 33 Ciencias Tecnológicas
AB Ocular artifacts (OA) are the most common artifacts in electroencephalography (EEG), significantly affectingsignal quality and analysis. Common approaches like indepentent component analysis (ICA) or regression-basedmethods address this problem but require several minutes of subject-specific EEG and electrooculography(EOG) calibration, making them impractical for real-time applications like brain–computer interfaces (BCI).In this study, we introduce EEGOAR-Net, a deep learning architecture designed to reduce OA in EEG. Itaddress these issues while also providing flexibility across various EEG montages. Based on U-Net architecture,EEGOAR-Net was trained with contaminated EEG signals in order to reconstruct them with OA attenuated,using SGEYESUB as the reference method. In addition, a novel training methodology based on masking signalsfrom different channels was applied to make EEGOAR-Net independent of the EEG montage used. A cross-validation analysis was conducted to assess EEGOAR-Net’s performance, demonstrating its ability to reduceEEG-EOG correlations to chance levels across most brain regions with minimal information loss. Thus, theperformance of EEGOAR-Net is comparable to that of the reference method without the need for subject-specific calibration or EOG channels. Furthermore, validation on an additional dataset confirmed effectiveblink reduction and superior preservation of neural information compared to the state-of-the-art models: 1D-ResCNN and IC-U-Net. EEGOAR-Net’s performance across datasets and versatility across montages prove it tobe a reliable and practical solution for EEG-based research and BCI applications, ensuring a notable reductionof OA on signal while maintaining the integrity of neural information.
PB Elsevier
SN 1746-8094
YR 2025
FD 2025
LK https://uvadoc.uva.es/handle/10324/78224
UL https://uvadoc.uva.es/handle/10324/78224
LA eng
NO Biomedical Signal Processing and Control, 2025, vol.110, p. 108147
NO Producción Científica
DS UVaDOC
RD 11-nov-2025