human epilepsy; neuronal networks; electroencephalogram; complex systems; networks; therapy; EEG
Naro Daniel, Rummel Christian, Schindler Kaspar, Andrzejak Ralph G (2014), Detecting determinism with improved sensitivity in time series: rank-based nonlinear predictability score., in Physical review. E, Statistical, nonlinear, and soft matter physics
, 90(3), 032913-032913.
Rummel Christian, Goodfellow Marc, Gast Heidemarie, Hauf Martinus, Amor Frédérique, Stibal Alexander, Mariani Luigi, Wiest Roland, Schindler Kaspar (2013), A Systems-Level Approach to Human Epileptic Seizures, in Neuroinformatics
, 11(2), 159-173.
Rummel Christian, Müller Markus, Hauf Martinus, Wiest Roland, Schindler Kaspar (2013), Dynamics of linear and nonlinear interrelation networks in peri-ictal intracranial EEG: seizure onset and termination, in Tetzlaff Roland (ed.), 162-174.
Jiruska Premek, de Curtis Marco, Jefferys John, Schewon Catherine, Schiff Steve, Schindler Kaspar (2013), Synchronization and Desynchronization in Epilepsy: Controversies and Hypotheses., in Journal of Physiology
, 15(591), 787-797.
Adrzejak Ralph, Schindler Kaspar, Rummel Christian (2012), Nonrandomness, nonlinear dependence, and nonstationarity of electroencephalographic recordings from epilepsy patients, in Phys. Rev. E
, 86(4), 046206-17p.
Schindler K Gast H Goodfellow M Rummel C (2012), On seeing the trees and the forest: Single-signal and multisignal analysis of periictal intracranial EEG., in Epilepsia
, 53(9), 1658-1668.
Gast Heidermarie, Schindler Kaspar, Rummel Christian, Herrmann Uli S., Roth Corinne, Hess Christian W., Mathis Johannes (2011), EEG correlation and power during maintenance of wakefulness test after sleep-deprivation., in Clinical Neurophysiology
, 122(10), 2025-2031.
Müller Markus, Baier Gerold, Jimenez Yurytzy Lopez, Marin Garcia Arlex, Rummel Christian, Schindler Kaspar (2011), Evolution of Genuine Cross-Correlation Strength of Focal Onset Seizures, in Journal of Clinical Neurophysiology
, 28(5), 450-462.
Schindler Kaspar, Gast Heidemarie, Stieglitz Lennart, Stibal Alexander, Hauf Martinus, Wiest Roland, Mariani Luigi, Rummel Christian (2011), Forbidden ordinal patterns of periictal intracranial EEG indicate deterministic dynamics in human epileptic seizures., in Epilepsia
, 52(10), 1771-1780.
Rummel Christian, Abela Eugenio, Müller Markus, Hauf Martinus, Scheidegger Olivier, Wiest Roland, Schindler Kaspar (2011), Uniform approach to linear and nonlinear interrelation patterns in multivariate time series, in Physical Review E
, 83, 066215-13.
Rummel Christian, Müller Markus, Baier Gerold, Amor Frédérique, Schindler Kaspar (2010), Analyzing spatio-temporal patterns of genuine cross-correlations, in Journal of Neuroscience Methods
, 191(1), 094-100.
Rummel Christian, Gast Heidemarie, Schindler Kaspar, Müller Markus, Amor Frédérique, Hess Christian, Mathis Johannes (2010), Assessing periodicity of periodic leg movements during sleep, in Frontiers in Neuroscience
, 4, 1-14.
Schindler Kaspar, Rummel Christian, Gast Heidemarie, Amor Frédérique, Stibal Alexander, Stieglitz Lennart, Wiest Roland, Hauf Martinus, Mathis Johannes, Meyer Klaus, Rüegg Stefan, Mariani Luigi, Raabe Andreas, Hess Christian W. (2010), Frontallappenepilepsie – dorsolateral, orbital, polar, in Epileptologie
, 27, 124-132.
Schindler Kaspar, Rummel Christian, Gast Heidemarie, Amor Frédérique, Stibal Alexander, Wiest Roland, Hauf Martinus, Mathis Johannes, Rüegg Stefan, Mariani Luigi, Raabe Andreas, Hess Christian W. (2009), Das Elektroenzephalogramm (EEG) in der prä-chirurgischen Epilepsiediagnostik, in EpilepsieNewsletter
, xxx(Juni 2010), 1-8.
Schindler Kaspar, Amor Frédérique, Gast Heidemarie, Müller Markus, Stibal Alexander, Mariani Luigi, Rummel Christian (2009), Peri-ictal correlation dynamics of high-frequency (80-200 Hz) intracranial EEG., in Epilepsy Research
, 89(1), 72-81.
Gast Heidemarie, Müller Markus, Rummel Christian, Roth Corinne, Mathis Johannes, Schindler Kaspar, Bassetti Claudio, Epileptic seizures as condensed sleep: an analysis of network dynamics from elecetroencephalogram signals, in Journal of Sleep Research
ContextEpilepsy is the second most common neurological disease. Recurrent epileptic seizures are associated with increased mortality, high morbidity, and a severe psycho-social burden for the patient and his/her family. Therefore the goal of epilepsy treatment is to render the patient seizure-free. However, since this goal is not achieved in one fourth of all patients, it is important to develop better therapies. To allow for a rational guidance of the development of improved therapies, a better understanding of the pathological changes of human epileptic brains is indispensable. In recent years, neuroimaging has provided important insights into structural changes associated with epilepsy and basic neuroscience has unravelled cellular and sub-cellular epileptogenic mechanisms. However, pathophysiological functioning is often not derivable from altered structure. Moreover, epileptic seizures are not produced by single neurons only, but result from excessive and aberrant collective activity and dynamic interaction of a very large number of nerve cells. Therefore it is important to study not only structure, but also function and to not only focus on single cells, but to strive to better understand large neural networks in the epileptic brain.General ObjectiveThe general objective of our study is to understand functional networks in human epileptic brains as a prerequisite for the rational development of more effective therapies. Our general study objective represents a specific example of one of the most fundamental problems of neuroscience, namely how segregated neuronal activity is integrated and coordinated, or, to widen the context even further, to the classical mereological problem of how parts are related to the whole. The understanding of this relationship, which is often discussed in an abstract way, may have very practical consequences in the present context of guiding the development of new therapies, because parts of the brain may be changed by surgical interventions, while the brain as a whole is affected by systemically applied drugs.MethodsIn the context of functional networks, the term "function" refers to neural electrical activity, which is recorded by the electroencephalogram (EEG). We will analyze multi-channel EEG recordings using quantitative methods originally developed in physics for the study of complex systems that consist of many interacting elements. A crucial characteristic of these methods is that they allow to study interdependencies on different scales. In other words, these methods allow to focus on parts of the functional networks as well as on the emergent activity produced by all the interacting functional networks as a whole. Specifically, mathematical tools from nuclear physics ("Random-Matrix Theory") and from graph theory will be applied, adjusted and further developed. These methods allow to detect dynamically formed "clusters", i.e. in our case those subsets of EEG channels that inter-depend more strongly with each other than with the rest. Furthermore, those EEG channels with an increased number of strong functional interactions with remote brain areas can be identified. Within the framework of graph theory such channels correspond to so-called "hubs", i.e. parts of a functional network that are specifically important to sustain and coordinate its activity. Linear (equal-time cross-correlation and finite-lag correlation) and nonlinear (mutual information) measures will be used to characterize the dynamic interdependencies between EEG channels. Rather than on bivariate relations between pairs of channels, emphasis will be put on multivariate relations of the entire set of channels. To this end, properties of the eigenvalues and eigenvectors of interdependence matrices as well as graph theoretical measures will be studied in this tripartite project. First, in order to assess the multivariate effect of lag correlations between signals recorded from different sites of the brain, the concept of the equal-time cross-correlation matrix will be generalized to finite time lags, in a way that conserves the practically important mathematical property of real and non-negative eigenvalues. Comparing the equal-time and the finite-lag correlation structure will allow to assess propagation of activity through functional networks. Second, to be sensitive not only to linear but also to nonlinear interdependencies, we will define functional networks based on the mutual information between EEG channels. Characteristics defined within the framework of Random Matrix Theory will be investigated for the multivariate generalization of mutual information. Based on preliminary results we expect that the nonlinear interdependencies are more sensitive to changes within the functional networks in epileptic brains. As a third part of the project, graph theoretical measures will be applied in a time resolved manner to interdependence matrices constructed from EEG recordings on the basis of equal-time cross-correlation, finite-lag correlation and mutual information. As a central methodological point, the results will be constantly interpreted in the context of clinical neurophysiology. Using the EEG data base of the Clinic of Neurology of the Inselspital in Bern, the methods can thus be evaluated for their practical utility, which will influence their further development.Clinical QuestionsThe following clinically important questions will be addressed:1. Is it possible to identify local parts ("clusters" and/or "hubs") within the functional networks in epileptic brains that show specific activity in the interictal and during different ictal stages, i.e. before and at seizure onset, during seizure propagation, before, at and following seizure termination?2. How does the emergent global activity of the functional networks as a whole differ between interictal and ictal stages?3. How do functional neuronal networks differ between patients suffering from focal-onset seizures vs. those suffering from primary generalized seizures?4. Are the functional networks of epileptic brains different in the interictal stages in those patients who have rare (or no more) seizures compared to those who have frequent seizures?5. How do anti-convulsive drugs act on the functional networks?6. Do brain regions, that have been surgically removed in patients who are post-operatively seizure-free, correspond to specific parts of functional networks (i.e. "clusters" and/or "hubs")?RelevanceA better understanding of functional networks in human epileptic brains is highly relevant in regard to improving existing or developing new therapeutic approaches. This study will contribute to this goal in at least a twofold way: First, signal analysis tools to assess neuronal network activity will be advanced, and second, the analysis results will be interpreted in the context of clinically relevant questions. The detection of clusters/hubs and the understanding of their relationship to the activity of the epileptic brain as a whole may have direct practical consequences. For example the goal of epilepsy surgery is to remove the so-called "epileptogenic zone", which often appears structurally normal. Especially in these cases it would be highly relevant to be able to identify those parts of the neuronal networks that are crucial for seizure generation based on EEG analysis. In addition, our study may be helpful by identifying also those parts of the functional networks that are important for seizure termination and may potentially be stabilized by modern tools such as electric brain stimulation. The characteristics of the emergent global activity of functional networks are highly important in regard to understanding the effects of anti-convulsive drugs. Better understanding of how drugs affect global network activity would be specifically relevant for assessing the anti-convulsive potential of new drugs.