We propose to estimate transfer entropy using a technique of symbolization. We demonstrate numerically that symbolic transfer entropy is a robust and computationally fast method to quantify the dominating direction of information flow between time series from structurally identical and nonidentical coupled systems. Analyzing multiday, multichannel electroencephalographic recordings from 15 epilepsy patients our approach allowed us to reliably identify the hemisphere containing the epileptic focus without observing actual seizure activity.

In both humans and animals, an insult to the brain can lead, after a variable latent period, to the appearance of spontaneous epileptic seizures that persist for life. The underlying processes, collectively referred to as epileptogenesis, include multiple structural and functional neuronal alterations. We have identified the T-type Ca 2ϩ channel Ca v 3.2 as a central player in epileptogenesis. We show that a transient and selective upregulation of Ca v 3.2 subunits on the mRNA and protein levels after status epilepticus causes an increase in cellular T-type Ca 2ϩ currents and a transitional increase in intrinsic burst firing. These functional changes are absent in mice lacking Ca v 3.2 subunits. Intriguingly, the development of neuropathological hallmarks of chronic epilepsy, such as subfield-specific neuron loss in the hippocampal formation and mossy fiber sprouting, was virtually completely absent in Ca v 3.2 Ϫ/Ϫ mice. In addition, the appearance of spontaneous seizures was dramatically reduced in these mice. Together, these data establish transcriptional induction of Ca v 3.2 as a critical step in epileptogenesis and neuronal vulnerability.

No abstract

We investigate the applicability of the permutation entropy H and a synchronization index γ that is based on the changing tendency of temporal permutation entropies to analyze noisy time series from nonstationary dynamical systems with poorly understood properties. Using model systems, we first study the interdependencies of parameters involved in the calculation of both measures. Having identified appropriate parameter settings we then analyze long-lasting EEG time series recorded from an epilepsy patient. Our findings indicate that γ could be of interest for studies on the predictability of epileptic seizures.

We investigate the relative merit of different linear and nonlinear synchronization measures for a characterization of the spatio-temporal dynamics of the epileptic process. Analyzing long-lasting multichannel electroencephalographic recordings from more than 20 epilepsy patients we show that all measures are able to identify brain regions of pathological synchronization associated with epilepsy, even during the seizure-free interval, and are able to detect a long-lasting transitional preseizure state. These findings render synchronization measures attractive for future prospective studies on seizure prediction.