Spatiotemporal dynamics of human epileptic seizures

1 In: Proceedings of the 3rd Experimental Chaos Conference, eds. R.G. Harrison et al., World Scientific, Singapore, pp. 26-30,1996

Spatiotemporal dynamics of human epileptic seizures Leonidas D. Iasemidis1,2,3, Konstantinos E. Pappas3, Jose C. Principe3, J. Chris Sackellares1,2,4 Departments of Neurology1, Neuroscience2, Electrical Engineering3 University of Florida, Gainesville, FL 32610-0236 Neurology Service4, Veterans Affairs Medical Center, Gainesville, FL 32608-1197 E-mail: leon @epilepsy.health.ufl.edu

Epilepsy is often called a window to the brain because it offers a unique opportunity for scientists to study and understand the function of the circuitry of the human brain during a fairly well defined and relatively simpler state, the epileptic seizure. We have previously shown that electrical activity recorded from cerebral cortex and generated by human epileptogenic foci exhibits certain mathematical characteristics of chaotic signals. We will present results that show that the epileptic seizure unfolds as a transition from spatiotemporal chaos to spatial order and temporal chaos, to spatiotemporal chaos, and that the key factor of this transition is a slow entrainment process among cortical and subcortical structures beginning minutes prior to a seizure manifestation. This spatiotemporal entrainment begins locally within cortical and hippocampal structures with subsequent communication between themselves.

1. Background Epilepsy, the recurrence of sudden disturbances of the brain function. is a common disorder affecting approximately 1% of the general population. These disturbances (seizures) result in a variety of intermittent clinical phenomena including motor, sensory, affective, cognitive, autonomic and psychic symptomology. In human epilepsy of mesial temporal lobe origin, seizures begin in the hippocampus and are often propagated throughout the brain. A central feature of the epileptogenic hippocampus is the tendency to make abrupt transitions to well organized oscillations, characteristic of a seizure. The temporal and orbitofrontal cortex also appear to play a role in the onset and spread of these seizures (Niedermeyer, 1987). Detection of potentially epileptogenic foci from the electroencephalogram (EEG) of epileptic patients depends upon the recognition of characteristic transient waveforms, such as spikes and sharp waves, in the ictal (during a seizure) or interictal (between seizures) periods (Gotman et al., 1985). Research along conventional lines of analysis has unsuccessfully sought to detect precursors of the epileptic seizure (Basar et al., 1989). While such approaches have generated descriptive information regarding interictal and ictal epileptogenic physiological phenomena, they have not been able to explain why seizures occur when they do and why they are terminated. From a dynamical perspective the onset of a seizure represents an abrupt phase transition from a complex to a less complex (more ordered) state. The spontaneous formation of organized spatial, temporal or spatiotemporal patterns in various physical, chemical or biological systems is common (Marcus et al., 1988). The critical element in such abrupt changes of the state of a deterministic system is the nonlinear nature of the system. The recorded electrical activity of the brain is a complex signal, the statistical properties of which depend on both time and space (Lopes da Silva, 1991). Characteristics of the EEG, such as the existence of limit cycles (Iasemidis, 1988), instances of bursting behavior, hysteresis, and amplitude dependent frequency behavior are among the long catalog of typical properties of nonlinear systems. By applying techniques from nonlinear dynamics, several researchers have provided evidence that the EEG is a nonlinear signal (for a review see Elbert et al., 1994 and references therein). Our group, employing analytic techniques developed for the analysis of complex nonlinear systems, was the first to demonstrate and quantify specific spatiotemporal dynamical changes in the electrical activity recorded from the surface of the cortex (ECoG) that begin several minutes before and end several minutes after an epileptic seizure. These changes appeared to evolve in a characteristic pattern, culminating in a seizure (Iasemidis et al., 1991). Our current studies have employed electrographic recordings from implanted electrodes placed bilaterally in the hippocampus and over the inferior temporal and orbitofrontal cortex in candidates for surgical excision of the seizure focus (Spencer et al., 1992).

2 2. Results A formal quantitative analysis of the preictal, ictal and postictal states was undertaken. Dynamical measures of the EEG signals like the correlation dimension ν and the maximum Lyapunov exponent Lmax were calculated, the former characterizing the complexity and the latter the unpredictability (chaoticity) of the data. The correlation dimension for ictal EEG data has been found to be a noninteger between 2.5 and 3.5 (Iasemidis et al., 1988). Based on this result, at least a 6 to 7 dimensional phase space is necessary to capture the details of the evolution of the system with time during the ictal state. The Lmax was calculated from sequential nonoverlapping EEG epochs of approximately 10 second duration, usually beginning 1 hour before and ending almost 1 hour after a seizure. The method of Wolf and co-workers (1985) for calculating (Lmax) was modified to deal with nonstationarities in the EEG, such as the interictal epileptogenic spike (Iasemidis et al., 1990).

Figure 1: Lyapunov values from ipsilateral orbitofrontal (black line) and contralateral orbitofrontal (grey line) electrodes over time.

Preliminary results about the dynamical interactions between the epileptogenic hippocampus and selected regions of the ipsilateral and contralateral frontal and temporal cortex have been generated. As with the ECoG recordings, preictal fluctuations in the value of Lmax over time are observed at each recording site (see Figures 1 and 2); note that all Lmax values are positive (chaos). Initially, these fluctuations are out of phase and differ in their mean values (spatiotemporal chaos). Beginning several minutes prior to the seizure, the mean values of Lmax at these sites start to converge. The fluctuations of Lmax become phase locked later, well before the seizure's onset. Just prior to the seizure, there is a widespread phase locking among areas of the cortical regions on both sides of the brain. Seconds after seizure's onset, the still positive Lmax values from all sampled regions of the brain, are phase locked in an abrupt transition to a more ordered state, the seizure (spatial order but still temporal chaos). After seizure's end, the preictally entrained brain regions start to unlock. Dynamical analysis of the electrical activity from the epileptogenic and non-epileptogenic hippocampus revealed different dynamical properties during the preictal, ictal, and postictal states (see Figure 3). Preictal values of Lmax in the epileptogenic hippocampus are lower than the ones of the normal contralateral hippocampus. This observation suggests that the epileptogenic hippocampus possesses fewer degrees of freedom than the more normal hippocampus. The two hippocampi are not entrained until seconds prior to or after the seizure's onset. The seizure's onset is first manifested in the epileptogenic hippocampus. Immediately after the end of the seizure, the of the epileptogenic hippocampus reaches similar values to the ones of the more normal side. However, the postictal entrainment. resulting from the occurrence of the seizure does not last longer than 5 minutes. Consequently, the epileptogenic hippocampus returns to its lower Lmax values, an indication of its inability to process information, or to operate as a controller, as effectively as the contralateral normal hippocampus does.

Figure 2: Lyapunov values from two electrodes (black and grey lines) in the ipsilateral inferior temporal cortex over time.

3

Figure 3: Lyapunov values from epileptogenic hippocampus (black line) and the contralateral hippocampus (grey line) over time.

3. Conclusion Preliminary studies of the human epileptic brain with depth EEG data have now confirmed our previous results with ECoG data. These studies suggest that a seizure is a spatiotemporal transition and occurs only when a sufficiently large area of the cerebrum becomes entrained dynamically. Although these observations constitute a first major step towards the characterization of the brain's transition to an epileptic seizure, several questions remain. For example, we still have only limited information regarding the nature of the dynamical processes that contribute to this entrainment. Being able to detect such an entrainment well before onset of an epileptic seizure, one may intervene and prevent the impending seizure well before its actual onset. Acknowledgments This work is supported by NIH Grant RO1 NS31451

References 1. Basar E., Bullock T.H., Brain dynamics: Progress and prospectives (Springer-Verlag, Heidelberg, 1989). 2. Elbert T., Ray W.J., Kowalik J., Skinner J.E., Graf K.E., Birbaumer N., Physiol. Rev. 74, 1-47 (1994). 3. Gotman J., Ives J.R., Gloor P., Long-term monitoring in epilepsy, (Elsevier, Amsterdam, 1985). 4. Iasemidis L.D., Sackellares J.C., Measuring chaos in the human brain, eds. Duke D.W., Pritchard W.S. (World Scientific, Singapore, 1991), p. 49-82. 5. Iasemidis L.D., Sackellares J.C., Zaveri H.P., Williams W.J., Brain Topogr., 2, 187-201 (1990). 6. Iasemidis L.D., Zaveri H.P., Sackellares J.C., Williams W.J., 25th Annual Rocky Mountain Bioengineering Symposium, 24, 187-193 (1988). 7. Lopes da Silva F., Electroenceph. Clin. Neurophysiol., 79, 81-93 (1991). 8. Marcus M., Aller S.M., Nicolis G., From chemical to biological organization, (Springer-Verlag, Berlin, 1988). 9. Niedermeyer E., Electroencephalography: Basic principles, clinical applications and related fields, eds. Niedermeyer E., Lopes da Silva F. (Urban Schwarzenberg, Baltimore, 1987), p. 593-617. 10. Spencer D.D., Pappas C.T.E., Clinical Neurosurgery, 38, 548-566 (1992). 11. Wolf A., Swift J.B., Swinney H.L., Vastano J.A., Physica D, 16, 285-317 (1985).

4