Visualization and modelling of STLmax topographic brain activity maps.

This paper evaluates the descriptive power of brain topography based on a dynamical parameter, the Short-Term Maximum Lyapunov Exponent (STLmax), estimated from EEG, for finding out a relationship of STLmax spatial distribution with the onset zone and with the mechanisms leading to epileptic seizures. Our preliminary work showed that visual assessment of STLmax topography exhibited a link with the location of seizure onset zone. The objective of the present work is to model the spatial distribution of STLmax in order to automatically extract these features from the maps. One-hour preictal segments from four long-term continuous EEG recordings (two scalp and two intracranial) were processed and the corresponding STLmax profiles were estimated. The spatial STLmax maps were modelled by a combination of two Gaussians functions. The parameters of the fitted model allow automatic extraction of quantitative information about the spatial distribution of STLmax: the EEG signal recorded from the brain region where seizures originate exhibited low-STLmax levels, long before the seizure onset, in 3 out of 4 patients (1 out of 2 of scalp patients and 2 out of 2 in intracranial patients). Topographic maps extracted directly from the EEG power did not provide useful information about the location, therefore we conclude that the analysis so far carried out suggests the possibility of using a model of STLmax topography as a tool for monitoring the evolution of epileptic brain dynamics. In the future, a more elaborate approach will be investigated in order to improve the specificity of the method.

Clustering approach to quantify long-term spatio-temporal interactions in epileptic intracranial electroencephalography.

Abnormal dynamical coupling between brain structures is believed to be primarily responsible for the generation of epileptic seizures and their propagation. In this study, we attempt to identify the spatio-temporal interactions of an epileptic brain using a previously proposed nonlinear dependency measure. Using a clustering model, we determine the average spatial mappings in an epileptic brain at different stages of a complex partial seizure. Results involving 8 seizures from 2 epileptic patients suggest that there may be a fixed pattern associated with regional spatio-temporal dynamics during the interictal to pre-post-ictal transition.