Abstracts

Rationale: The misconception that rhythmic seizure activity is associated with lower complexity than background often arise from a lack of definition of complexity itself. Often used as a general concept complexity is not always clearly defined. We previously introduced a measure of signal complexity based on the matching pursuit decomposition. We now present a more systematic analysis of intracranial recordings of epileptic seizures to study changes in complexity associated with ictal events. Methods: Seizures recorded from 20 consecutive patients with intracranial recordings arrays (subdural grids with strips) were analyzed for this study. Patient ages ranged from 10 to 50 years (27±13 yo). Eleven patients were diagnosed with mesial temporal lobe epilepsy, seven with neocortical onset epilepsy and two were found to have bilateral onset seizures. All epileptic events were analyzed with the matching pursuit (4.096 sec window; 1.024 sec overlapping) time-frequency decomposition method using the channel closest to the seizure focus. Gabor atom density (GAD) was computed for all seizures recorded and identified by the clinicians. These events included sub-clinical seizures and partial seizures (simple or complex) with or without secondary generalization. Results: GAD complexity was computed for 178 seizures from the 20 patients included in this analysis. GAD increases during seizure averages 104+/-55% overall. The lowest increases are observed for extremely focal simple partial seizures with neocortical onset. Increased signal complexity early in the seizure is dependent of the type on onset pattern. Electrodecremental onsets are the most variable as complexity in these instances depends on the underlying high frequency composition of the signal. Occurrence of such components increases the complexity of the signal but such high frequency activity not always identified. Even in seizures with initial electrodecremental patterns later seizure evolution is often associated with increased seizure complexity. Conclusions: Complexity as measured by GAD is a highly reliable method for identification of ictal events. In contrast to other detection methods, GAD does not need to be adjusted for particular patterns of seizures in a given patient to maximize sensitivity or specificity. The number of atoms measured influences the computational efficiency of the analysis. For intracranial ictal recordings this method permits retrospective review to identify subclinical or unmarked events in this important group of presurgical patients. Supported by grant NIH 48222.