Seizure prediction techniques: robustness and performance issues

Abstract

Recent research in the field of seizure prediction has shown that pre-seizure states lasting minutes to hours can be detected from analysis of EEG time series. This time frame would allow development and testing of specific seizure prevention techniques. Thus, there is now growing interest to bring these analysis techniques to a broader application. Similar demands come from the industry as there is strong interest to incorporate analysis techniques into EEG acquisition systems and, moreover, to develop miniaturized (implantable) devices for seizure prediction and possibly prevention.