Abstracts

Rationale: Resection of the Seizure Onset Zone (SOZ) represents a valuable treatment option for patients suffering from drug resistant epilepsy. During presurgical evaluation high resolution imaging, PET, psychological testing, and long term EEG recordings over several days are examined to determine the SOZ. Further improvements concerning the localization quality of the SOZ can be achieved by applying novel source localization methods to scalp EEG signals. Over the last two decades a lot of research on this topic has been done, which is summarized in comprehensive reviews [1]. Today, locations of neuronal sources corresponding to certain EEG sequences can be calculated using commercially available software tools [2]. From our point of view these tools have two major drawbacks: 1.) Too many user-interactions are necessary to reach reasonable results, and 2.) in most cases only interictal spikes are investigated [1]. In order to overcome these problems, we propose an automatic SOZ detection system focusing on ictal EEG only. Methods: The proposed approach utilizes the AIT seizure detection system EpiScan [3] and extends it by a frequency domain source localization module for SOZ detection. It has been shown that EpiScan detects rhythmic epileptiform seizure activity, which is often present during the early phase of epileptic seizures, with a very high sensitivity and spezifity. This detected rhythmic activity of the early ictal EEG signal is the input for the SOZ localization. A frequency domain version of the Eigenspace Beamformer (ESB) is used to estimate the locations and the strengths of the neuronal sources corresponding to the detected rhythmic activity. We define the centre of gravity of the resulting distribution of the neural activity within the brain as the SOZ.Results: The first results of our proposed automatic SOZ detector are based on long-term EEG recordings of 2 patients who underwent a presurgical workup due to drug resistant epilepsy. The EpiScan system detected 83% of the seizures (5 out of 6 seizures, 3 seizures per patient). Using the onsets of the rhythmic patterns of the ictal EEG for source localization, the SOZs of all detected seizures are in excellent accordance with the provided medical report. Conclusions: Our first results for automatic SOZ detection are encouraging. In all cases the source localization results are in good accordance with clinical findings. This proof of concept shows that it is possible to do automatic SOZ detection based on the early seizure activity, which will increase the applicability of source localization in everyday clinical practice. [1] C. M. Michel, M. M. Murray, G. Lantz, S. Gonzalez, L. Spinelli, und R. Grave de Peralta, EEG source imaging, Clinical Neurophysiology: Bd. 115, Nr. 10, S. 2195-2222, Okt. 2004. [2] http://www.neuroscan.com/curry5.cfm. [3] T. Kluge, M. M. Hartmann, C. Baumgartner, und H. Perko, Automatic Detection of Epileptic Seizures in scalp EEG-Recordings Based on Subspace Projections, Epilepsia, Bd. 50, S. 26-27, 2009.