1471-2202-13-S1-P95 1471-2202 Poster presentation <p>Interpretation of seizure evolution pathways via a mean-field cortical model</p> DadokMVeravdadok@berkeley.edu SzeriJ Andrew KirschHeidi SleighJamie LopourBeth

Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA

Center for Neural Engineering and Prostheses, UC Berkeley and UC San Francisco, CA, USA

School of Medicine, University of California, San Francisco, CA, 94143, USA

School of Medicine, University of Auckland, Grafton, Auckland, 1142, New Zealand

Department of Neurobiology, University of California, Los Angeles, CA, 90095, USA

 BMC Neuroscience <p>Twenty First Annual Computational Neuroscience Meeting: CNS*2012</p>Jean-Marc Fellous and Astrid PrinzMeeting abstracts<p>Twenty First Annual Computational Neuroscience Meeting: CNS*2012</p>Decatur, GA, USA21-26 July 2012http://www.cnsorg.org/cns-2012-atlantadecatur1471-2202 2012 13 Suppl 1 P95 http://www.biomedcentral.com/1471-2202/13/S1/P95 10.1186/1471-2202-13-S1-P95 1672012 2012Dadok et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Treatment of epilepsy is a challenging task. Difficulties arise in choosing the best pharmaceutical drugs for specific patients, and if limited benefits result, attempting to choose the best alternative treatment. It is not always clear why a treatment that works well for one patient does not work well for another. Better understanding of the epileptic brain, such as differentiating between possible biological mechanisms driving seizure evolution, may offer insight into these problems.

This work makes use of a dynamical model of the human cortex based on the underlying physiology and local anatomy of the brain 1 . This mean-field model can generate electrocorticogram-like (ECoG-like) data in both normal states and seizing states under different plausible parameter configurations 2 3 . There are several different sets of parameters which, when changed, drive the model from a non-seizing state into a seizing state and back. An example of one such pathway is displayed in Figure 1. The presence of different parameter sets that can trigger seizures or exit seizures when varied indicates that there are different biological pathways to and from seizure, and these pathways can be described in terms of the fundamental physical quantities in the model 2 . This work will demonstrate a method that takes ECoG-like data from a model cortex evolving into seizure and identifies which parameters are changing—thus identifying which associated physiological mechanisms may be leading the brain into a seizure. In conjunction with other tools, this method will leverage feature selection and dimensionality reduction algorithms. The same technique may be applied to experimentally collected ECoG seizure data as well.

<p>Figure 1</p>

Illustrates a possible path via two model parameters varying and the resultant ECoG-like data. The exact definitions of all parameters are left to the poster. In this illustrative figure parameter 1 is related to the number of inhibitory synaptic connections and parameter 2 is related to the strength of the subcortical input on excitatory neurons in the cortex. A. Locations in parameter space. Shading relates to the permutation complexity of the signal. B. ECoG-like data signals generated by the model corresponding to locations in parameter space. At locations 1 and 2, the simulated signal through time shows normal cortex activity. At location 3, the ECoG-like data signal through time is in a seizure state with more regular sustained oscillations.

Illustrates a possible path via two model parameters varying and the resultant ECoG-like data. The exact definitions of all parameters are left to the poster. In this illustrative figure parameter 1 is related to the number of inhibitory synaptic connections and parameter 2 is related to the strength of the subcortical input on excitatory neurons in the cortex. A. Locations in parameter space. Shading relates to the permutation complexity of the signal. B. ECoG-like data signals generated by the model corresponding to locations in parameter space. At locations 1 and 2, the simulated signal through time shows normal cortex activity. At location 3, the ECoG-like data signal through time is in a seizure state with more regular sustained oscillations.

Acknowledgements

This work was partly supported by a research grant from the National Science Foundation and an NSF Graduate Research Fellowship to V. Dadok.

 <p>Theoretical predictions for spatial covariance of the electroencephalographic signal during the anesthetic-induced phase transition: Increased correlation length and emergence of spatial self-organization</p>Steyn-RossMLSteyn-RossDASleighJWWhitingDRPhys Rev E200368021902-102190218<p>Mechanisms of seizure propagation in a cortical model</p>KramerMASzeriAJSleighJWKirschHEJ. Comput Neurosci200722638010.1007/s10827-006-9508-516998642<p>A continuous mapping of sleep states through association of EEG with a mesoscale cortical model</p>LoupourBATasogluSKirschHESleighJWSzeriAJJ. Comput Neurosci20113047148710.1007/s10827-010-0272-1305836820809258