Investigating the impact of the regularization parameter on EEG resting-state source reconstruction and functional connectivity using real and simulated data
F. Leone,
A. Caporali,
A. Pascarella,
C. Perciballi,
O. Maddaluno,
A. Basti,
P. Belardinelli,
L. Marzetti,
G. Di Lorenzo,
V. Betti
Affiliations
F. Leone
Department of Psychology, Sapienza University of Rome, via dei Marsi 78, Rome, 00185, Italy; IRCCS Fondazione Santa Lucia, via Ardeatina, 354, Rome, 00179, Italy; Corresponding authors.
A. Caporali
Faculty of Veterinary Medicine, University of Teramo, via R. Balzarini 1, Teramo, 64100, Italy,; International School of Advanced Studies, University of Camerino, via Gentile III Da Varano, Camerino, 62032, Italy
A. Pascarella
Institute for Computational Applications, CNR, Italy
C. Perciballi
Department of Psychology, Sapienza University of Rome, via dei Marsi 78, Rome, 00185, Italy; IRCCS Fondazione Santa Lucia, via Ardeatina, 354, Rome, 00179, Italy
O. Maddaluno
Department of Psychology, Sapienza University of Rome, via dei Marsi 78, Rome, 00185, Italy; IRCCS Fondazione Santa Lucia, via Ardeatina, 354, Rome, 00179, Italy
A. Basti
Department of Neuroscience, Imaging and Clinical Sciences, “G. d'Annunzio” University of Chieti-Pescara, via dei Vestini, Chieti, 66100, Italy
P. Belardinelli
CIMeC, Center for Mind/Brain Sciences, University of Trento, via delle Regole, 101, Mattarello-Trento, 38123, Italy
L. Marzetti
Department of Neuroscience, Imaging and Clinical Sciences, “G. d'Annunzio” University of Chieti-Pescara, via dei Vestini, Chieti, 66100, Italy; Institute for Advanced Biomedical Technologies, “G. d'Annunzio” University of Chieti-Pescara, via Luigi Polacchi, Chieti, 66100, Italy
G. Di Lorenzo
IRCCS Fondazione Santa Lucia, via Ardeatina, 354, Rome, 00179, Italy; Laboratory of Psychophysiology and Cognitive Neuroscience, University of Rome Tor Vergata, Rome, Italy
V. Betti
Department of Psychology, Sapienza University of Rome, via dei Marsi 78, Rome, 00185, Italy; IRCCS Fondazione Santa Lucia, via Ardeatina, 354, Rome, 00179, Italy; Corresponding authors.
Accurate EEG source localization is crucial for mapping resting-state network dynamics and it plays a key role in estimating source-level functional connectivity. However, EEG source estimation techniques encounter numerous methodological challenges, with a key one being the selection of the regularization parameter in minimum norm estimation. This choice is particularly intricate because the optimal amount of regularization for EEG source estimation may not align with the requirements of EEG connectivityanalysis, highlighting a nuanced trade-off. In this study, we employed a methodological approach to determine the optimal regularization coefficient that yields the most effective reconstruction outcomes across all simulations involving varying signal-to-noise ratios for synthetic EEG signals. To this aim, we considered three resting state networks: the Motor Network, the Visual Network, and the Dorsal Attention Network. The performance was assessed using three metrics, at different regularization parameters: the Region Localization Error, source extension, and source fragmentation. The results were validated using real functional connectivity data. We show that the best estimate of functional connectivity is obtained using 10−2, while 10−1 has to be preferred when source localization only is at target.
