000186311 001__ 186311
000186311 005__ 20210129214852.0
000186311 0247_ $$2doi$$a10.1016/j.neuroimage.2014.12.040
000186311 0247_ $$2WOS$$aWOS:000349618600033
000186311 0247_ $$2altmetric$$aaltmetric:3054250
000186311 0247_ $$2pmid$$apmid:25541187
000186311 037__ $$aFZJ-2015-00387
000186311 082__ $$a610
000186311 1001_ $$0P:(DE-Juel1)144558$$aEngemann, Denis$$b0$$eCorresponding Author$$ufzj
000186311 245__ $$aAutomated model selection in covariance estimation and spatial whitening of MEG and EEG signals.
000186311 260__ $$aOrlando, Fla.$$bAcademic Press$$c2015
000186311 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1423730017_13948
000186311 3367_ $$2DataCite$$aOutput Types/Journal article
000186311 3367_ $$00$$2EndNote$$aJournal Article
000186311 3367_ $$2BibTeX$$aARTICLE
000186311 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000186311 3367_ $$2DRIVER$$aarticle
000186311 520__ $$aMagnetoencephalography and electroencephalography (M/EEG) measure non-invasively the weak electromagnetic fields induced by post-synaptic neural currents. The estimation of the spatial covariance of the signals recorded on M/EEG sensors is a building block of modern data analysis pipelines. Such covariance estimates are used in brain–computer interfaces (BCI) systems, in nearly all source localization methods for spatial whitening as well as for data covariance estimation in beamformers. The rationale for such models is that the signals can be modeled by a zero mean Gaussian distribution. While maximizing the Gaussian likelihood seems natural, it leads to a covariance estimate known as empirical covariance (EC). It turns out that the EC is a poor estimate of the true covariance when the number of samples is small. To address this issue the estimation needs to be regularized. The most common approach downweights off-diagonal coefficients, while more advanced regularization methods are based on shrinkage techniques or generative models with low rank assumptions: probabilistic PCA (PPCA) and factor analysis (FA). Using cross-validation all of these models can be tuned and compared based on Gaussian likelihood computed on unseen data.We investigated these models on simulations, one electroencephalography (EEG) dataset as well as magnetoencephalography (MEG) datasets from the most common MEG systems. First, our results demonstrate that different models can be the best, depending on the number of samples, heterogeneity of sensor types and noise properties. Second, we show that the models tuned by cross-validation are superior to models with hand-selected regularization. Hence, we propose an automated solution to the often overlooked problem of covariance estimation of M/EEG signals. The relevance of the procedure is demonstrated here for spatial whitening and source localization of MEG signals.
000186311 536__ $$0G:(DE-HGF)POF3-572$$a572 - (Dys-)function and Plasticity (POF3-572)$$cPOF3-572$$fPOF III$$x0
000186311 7001_ $$0P:(DE-HGF)0$$aGramfort, A.$$b1
000186311 773__ $$0PERI:(DE-600)1471418-8$$a10.1016/j.neuroimage.2014.12.040$$p328–342$$tNeuroImage$$v108$$x1053-8119$$y2015
000186311 8564_ $$uhttps://juser.fz-juelich.de/record/186311/files/FZJ-2015-00387.pdf$$yRestricted
000186311 909CO $$ooai:juser.fz-juelich.de:186311$$pVDB
000186311 915__ $$0StatID:(DE-HGF)0100$$2StatID$$aJCR
000186311 915__ $$0StatID:(DE-HGF)0110$$2StatID$$aWoS$$bScience Citation Index
000186311 915__ $$0StatID:(DE-HGF)0111$$2StatID$$aWoS$$bScience Citation Index Expanded
000186311 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection
000186311 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bThomson Reuters Master Journal List
000186311 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS
000186311 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline
000186311 915__ $$0StatID:(DE-HGF)0310$$2StatID$$aDBCoverage$$bNCBI Molecular Biology Database
000186311 915__ $$0StatID:(DE-HGF)0420$$2StatID$$aNationallizenz
000186311 915__ $$0StatID:(DE-HGF)1030$$2StatID$$aDBCoverage$$bCurrent Contents - Life Sciences
000186311 915__ $$0StatID:(DE-HGF)1050$$2StatID$$aDBCoverage$$bBIOSIS Previews
000186311 915__ $$0StatID:(DE-HGF)9905$$2StatID$$aIF >= 5
000186311 9141_ $$y2015
000186311 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)144558$$aForschungszentrum Jülich GmbH$$b0$$kFZJ
000186311 9130_ $$0G:(DE-HGF)POF2-333$$1G:(DE-HGF)POF2-330$$2G:(DE-HGF)POF2-300$$aDE-HGF$$bGesundheit$$lFunktion und Dysfunktion des Nervensystems$$vPathophysiological Mechanisms of Neurological and Psychiatric Diseases$$x0
000186311 9130_ $$0G:(DE-HGF)POF2-89572$$1G:(DE-HGF)POF2-89570$$2G:(DE-HGF)POF3-890$$aDE-HGF$$bKey Technologies$$lDecoding the Human Brain$$v(Dys-)function and Plasticity$$x1
000186311 9131_ $$0G:(DE-HGF)POF3-572$$1G:(DE-HGF)POF3-570$$2G:(DE-HGF)POF3-500$$3G:(DE-HGF)POF3$$4G:(DE-HGF)POF$$aDE-HGF$$bKey Technologies$$lDecoding the Human Brain$$v(Dys-)function and Plasticity$$x0
000186311 9201_ $$0I:(DE-Juel1)INM-3-20090406$$kINM-3$$lKognitive Neurowissenschaften$$x0
000186311 980__ $$ajournal
000186311 980__ $$aVDB
000186311 980__ $$aI:(DE-Juel1)INM-3-20090406
000186311 980__ $$aUNRESTRICTED