Speaker

Prof. Leon Iasemidis

Title

"Prediction of Epileptic Seizures"

Abstract

Studies in patients with severe epilepsy suggest that a seizure begins long before its symptoms become apparent. The signs of an impending seizure can be picked up on an electroencephalogram from 10 minutes to hours before the seizure becomes obvious. A computational analysis of brain waveforms, which includesquadratic binary programming and embedding techniques from chaos theory, can be used to predict seizures, aid the identification of the seizure source and guide the excision of brain tissue in those patients who are candidates for surgery. A conjecture resulting from this analysis is that seizures may serve as resetting mechanisms of the brain and, contrary to current belief, it is necessary that they occur when they do. In particular, it appears like the progressive entrainment of normal brain sites with the epileptogenic zone prior to a seizure is disrupted after the seizure's occurence.