Event Abstract

NNMF connectivity microstates : A new approach to represent the dynamic brain coordination.

  • 1 Brain Innovation B.V., Netherlands
  • 2 Artificial Intelligence & Information Analysis Laboratory, Aristotle University, Department of Informatics, Greece
  • 3 Neuroinformatics Group, Aristotle University of Thessaloniki, www.neuroinformatics.gr, Greece
  • 4 School of Music Studies, Faculty of Fine Arts, Aristotle University, Greece

Symbolic dynamics (Rajagopalan et al. 2007) in the context of EEG are highly interwoven with the notion of Lehmann's microstates (Lehmann et al. 1987, 2010). These are detectable in the multichannel signal, from both event-related and spontaneous activity recordings, as recurrent quasi-stable scalp potential maps lasting from tens to hundreds of milliseconds. Discrete symbols can be assigned to these microstates, and since they are characterized by great consistency across subjects, this offers a very efficient data reduction method and further facilitates the derivation of brain dynamics descriptors (Ville et al. 2010; Dimitriadis et al., 2012). However, the whole approach relies on signal amplitude characteristics and, hence, ignores aspects of neural coordination that involves genuine interactions beyond coincidence in timing. Inspired by the simplicity and effectiveness of microstates, we have recently introduced an alternative representation that aims at analyzing functional-connectivity patterns (Dimitriadis et al. 2013, 2015). The relevant microstates, functional connectivity microstates (FCμstates), proved to be convenient descriptors for tracking inter-areal synchronization during cognitive ERP responses and revealed dynamical trends in a parsimonious fashion. We increase the descriptive power of the FCμstates approach by integrating a non-negative matrix factorization (NNMF) (Lee and Sung, 1999) algorithmic step. Here -for the first time- NNMF is applied to functional connectivity patterns, which are formed by inherently positively-valued measurements. The introduced approach starts by deriving time-indexed functional connectivity profiles based on pairwise, quasi-instantaneous, estimates of inter-areal phase synchronization. Time series of functional connectivity patterns are then formed (Lachaux et al. 1999). As a crucial step towards understanding the connectivity dynamics, non-negative matrix factorization (NNMF) is employed for data learning purposes. The extracted basis vectors are used for the re-parameterization of the functional connectivity within a ‘‘reduced’’ space. They correspond to the most essential ingredients for a parsimonious, part-based, representation of the original connectivity patterns. Within this low-dimensional, re-parameterized space, symbolization (by means of vector-quantization) (Martinetz et al. 1993) is performed and associated with the semantics of the brain's network organization. This steps transforms the sequence of connectivity patterns into a time series of symbols. Each symbol represents a homogeneous class of patterns, or equivalently, a connectivity microstate (FCμstate). The study of emerging symbolic dynamics can mediate the understanding of network-organization dynamics, since descriptors like Markov-Chains can now be applied in order to model complex neurodynamics phenomena. The whole procedure can be applied in different ways depending on the scope of each study. For instance, a subject-specific ‘alphabet’ of prototypical connectivity patterns can be designed and used to compare different recording conditions based on the recurrence of FCμstates. Alternatively, by adopting a collective design strategy (across a large population of subjects), commonalities in a population can be deduced that will reflect universal functional connectivity (re)organization trends. The proposed methodology was applied to multichannel EEG signals, to investigate how music influences the spatiotemporal network profile of ongoing brain activity. Brain activity at rest and during music listening was recorded from 14 volunteers. The favorite song of each participant and a 'neutral' piece of music, common among the participants, were employed to bring a subject's brain in two distinct states. Our design aimed at detecting music induced changes (by comparing resting state vs. music listening), providing indications about subject's engagement to music (by contrasting 'neutral' vs. 'favorite' music) and facilitating the detection of commonalities across subjects (by delivering a common music track). The included results point to the importance of particular brain rhythms in shaping connectivity during listening to music, the realization of distinctive functional-connectivity microstates (organization patterns) and the emergence of music-related metastability phenomena (Tognoli and Kelso, 2014). Based on class separability and statistical analysis of the symbolic time series, we identified two microstates the presence/absence of which signals the states of resting/musing-listening. Using the two mined prototypical connectivity patterns, we attempted a network characterization (Sporns, 2011) of the most distinctive FCstates. The most important difference between these two microstates was detected in the spatial layout of the connectivity strength.

References

Rajagopalan, V., Ray, A., Samsi, R., & Mayer, J. (2007). Pattern identification in dynamical systems via symbolic time series analysis. Pattern Recognition,40(11), 2897-2907.

Lehmann, D., Ozaki, H., & Pal, I. (1987). EEG alpha map series: brain micro-states by space-oriented adaptive segmentation. Electroencephalography And Clinical Neurophysiology, 67(3), 271-288.

Lehmann, D., Pascual-Marqui, R., Strik, W., & Koenig, T. (2010). Core networks for visual-concrete and abstract thought content: A brain electric microstate analysis. Neuroimage, 49(1), 1073-1079.

Van de Ville, D., Britz, J., and Michel, C. M. Eeg microstate sequences in healthy humans at rest reveal scale-free dynamics. Proceedings of the National Academy of Sciences 107, 42 (2010), 18179–18184.

Dimitriadis SI, Laskaris NA, Tsirka V, Erimaki S, Vourkas M, Micheloyannis S, Fotopoulos S.A novel symbolization scheme for multichannel recordings with emphasis on phase information and its application to differentiate EEG activity from different mental tasks. Cognitive Neurodynamics, Volume 6, Number 1, 8 February 2012, pp.107-113.

Dimitriadis SI, Laskaris NA, Tzelepi A. On the quantization of time-varying phase synchrony patterns into distinct Functional Connectivity Microstates (FCμstates) in a multi-trial visual ERP paradigm, Volume 26, Issue 3, July 2013, pp 397-409

Dimitriadis SI, Laskaris NA, Micheloyannis S.Transition Dynamics of EEG Network Microstates unmask developmental and task differences during mental arithmetic and resting wakefulness. Cognitive Neurodynamics August 2015, Volume 9, Issue 4, pp 371-387

Lee, D. D., and Seung, H. S. Learning the parts of objects by non-negative matrix factorization. Nature 401, 6755 (October 1999), 788–791.

Lachaux, J., Rodriguez, E., Martinerie, J., Varela, F., & others,. (1999). Measuring phase synchrony in brain signals. Human Brain Mapping, 8(4), 194-208.

Martinetz, T. M., Berkovich, S. G., & Schulten, K. J. (1993). Neural-gas' network for vector quantization and its application to time-series prediction. Neural Networks, IEEE Transactions on, 4(4), 558-569.

Tognoli E, Kelso JAS.The metastable brain. Neuron Volume 81, Issue 1, 8 January 2014, Pages 35–48
Sporns, O. Networks of the Brain. MIT press, 2011

Keywords: Multichannel EEG, Symbolic dynamics, phase synchrony, Time-Varying Connectivity Graphs, connectome, NMF

Conference: Neuroinformatics 2016, Reading, United Kingdom, 3 Sep - 4 Sep, 2016.

Presentation Type: Poster

Topic: General neuroinformatics

Citation: Marimpis AD, Dimitriadis SI, Adamos DA and Laskaris NA (2016). NNMF connectivity microstates : A new approach to represent the dynamic brain coordination.. Front. Neuroinform. Conference Abstract: Neuroinformatics 2016. doi: 10.3389/conf.fninf.2016.20.00022

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 25 Apr 2016; Published Online: 18 Jul 2016.

* Correspondence: Mr. Avraam D Marimpis, Brain Innovation B.V., Maastricht, Netherlands, avraam.marimpis@gmail.com

© 2007 - 2020 Frontiers Media S.A. All Rights Reserved