Time Series and Interactions: Data Processing in Epilepsy Research

Abstract

Computational methods can have significant contribution to the epilepsy research not only by modeling, but in the field of data analysis as well. The main aims of data analysis in epileptology are related to the possible treatment methods: Considering surgical treatment, the aim is to find any marker to localize the epileptic tissue, the seizure onset zone. Considering deep or inracranial stimulations, additional aims are to predict the seizures and the effect of stimulation. Regarding pharmacological treatment, the aim is to understand the roles of different channels to the cellular and network dynamics, and their possible modulation. The vast amount of neural data opened a new era of brain research where new data analysis methods are highly needed to take a full advantage of the resources we have. The traditional way of seizure onset zone localization uses the extreme good pattern matching or mismatch recognition skill of the human brain, and try to localize the appearance of the first pathological patterns at the initiation of the epileptic seizures. Pathological patterns typically differs in their frequency and wave­shape but can appear in different forms. This plurality of the wave­shapes, as well as the fast spreading of signals through intracortical pathways or the possibility of multiple site or deep origin makes the identification of the seizure onset zone ambiguous in many cases even for clearly recorded seizures. Thus, the challenge have different levels, outlining a roadmap for data analysis in epilepsy: ­ Identification of seizure onset zones based on clear, high density, subdural electrode recordings during the initiation of the epileptic seizure. ­ Identification of the seizure onset zones based on invasive recordings without epileptic seizures. ­ Identification of seizure onset zone with noninvasive methods. In this chapter, the detection and analysis methods for possible markers of the epileptic tissue, either ictal and interictal, are discussed. Interictal epileptic spikes Interictal epileptic spikes are the first candidate among the epilepsy related electrophysiological events promising, that seizure onset zone could be determined in the absence of ictal recordings. From data analysis point of view, interictal epileptic spikes are relatively easily detectable signals, due to their large amplitude and stereotype waveforms. The only major difficulty is caused by movement and other artifacts. However, it turned out, that occurrence of spikes extends well beyond the limits of the actual epileptic tissue, so as the spike intensity maps have limited determinative power on seizure onset zone. The reason of this much wider observability of the interictal spikes presumably lies on that, the majority of the cortical interictal spikes are evoked signals, evoked by synaptic activity arriving from a far located seizure onset zone. This observation raises the question, if there are any reliable morphological difference between spikes with local and remote origin. Reliable distinction between them, would make possible delimiting the seizure onset zone without recorded