Abstracts

Epilepsy is a brain disorder where neuronal networks in the brain behave abnormally. There is accumulating evidence that seizures may be [quot]predicted[quot] up to 10 minutes to several hours before the clinical manifestation of the seizure event. However, such prediction algorithms are not still sufficient for clinical applications. In this study, we propose a novel quantitative method, DFA ([italic]Detrended Fluctuation Analysis[/italic]), that can detect abnormalities in brain electrical activity. In addition, for comparison purposes the results of other promising techniques, synchronization and dynamical similarity index, as well as Hjorth parameters are applied to high bandwidth cortical EEG. The results show that our proposed technique (DFA) has a greater sensitivity to brain activity. DFA might be a useful methodology for focal antiepileptic drug administration or electrical stimulation, thereby enabling complete control of seizures without the need for surgical intervention., Intracranial EEG data recorded from grids and strips from 3 patients in the Epilepsy Unit at the Rainbow Babies and Children[apos]s within University Hospitals of Cleveland were used to evaluate the performance of the DFA algorithm. Patients were 51, 26 and 5 years of age. The adults had right temporal onset seizures (four and five seizures, respectively). The child had over 100 left hemisphere seizures. The data were recorded at a sampling rate of Samples/Sec. The length of the data for the first patient is 36 hours and for others more than 5 continuous days. To quantify brain activity the DFA algorithm, which is a technique for quantifying the long-range correlation characteristics in non-stationary physiological time-series data, was chosen. In addition to detect the slope of the log-log plot of DFA, an automated power law slope detection technique was used., Non overlapping 10 second data windows were chosen for the DFA analysis. The raw EEG patient data was used directly without any preprocessing or normalization. The slopes of the log-log plot of DFA during normal brain activity, transient state and seizure (ictal) were 0.98[plusmn]0.14, 1.42[plusmn]0.17, and 1.71[plusmn]0.12 respectively. While the change from the non-seizure to seizure state occurred suddenly, there were some seizures in which the brain activity increased as quantified by the slope of log-log DFA plot starting several minutes in advance of clinically- identified EEG seizure onset. Moreover, the brain activity as measured by the DFA can remain high up to 3-4 hours after the seizure event., DFA is a robust, sensitive, and not reference based analysis technique that can be used for the analysis of cortical EEG. Both sensitivity and robustness to noise make DFA a good candidate for further investigation as a methodology for seizure detection and prediction.,