A system for epileptic seizure focus detection based on EEG analysis
UCAmI'12 Proceedings of the 6th international conference on Ubiquitous Computing and Ambient Intelligence
Hi-index | 0.00 |
Several methods have been suggested to predict the onset of epileptic seizures from EEG data. We evaluated the performance of three predictions methods: the "dynamical similarity index", the "effective correlation dimension" and an extended, prospective version of the "accumulated energy". These prediction methods were applied on a large pool of intracranial EEG data from 21 patients. Altogether, 582 hours EEG data and 88 seizures were investigated.The "seizure prediction characteristic" was used as assessment criterion. It considers the strong dependency between sensitivity and the false prediction rate. For a rate of 1 to 3.6 false predictions per day, the similarity index yields a sensitivity between 21% and 42%, which was the best result of the three examined prediction methods. The extended version of the accumulated energy achieves a sensitivity between 18% and 31%, the effective correlation dimension between 13% and 30%.