Abstracts

Rationale: Five seizure detection schemes based on previously reported dynamical properties exhibited by EEG seizure data were employed in experimental rat model of spontaneous limbic seizures in order to evaluate their performance and suitability as a tool for a real-time seizure detection/intervention system. Methods: Fifty-day-old Sprague Dawley rats were stereotactically implanted in the right ventral hippocampus with a pair of 150 µ stainless steel wires. Rats were electrically stimulated by delivering a suprathreshold stimulus of 334 ± 159 μA to the implanted electrodes for 88 ± 26 minutes to induce status epilepticus (n=7). To monitor the presence and the severity of seizures, animals were continuously video/EEG monitored starting within the first two weeks post SE and seizure grades were determined using a modified Racine 0-5 seizure scale. Each animal is connected through a 6-channel commutator and shielded cable to the EEG recording system, which consists of an analog amplifier, a 12 bit A/D converter, and recording software (HARMONIE 5.2, Stellate Inc. Montreal), which is synchronized to a video unit for time-locked monitoring of behavioral changes. Each channel is sampled at a uniform rate of 200 Hz and filtered using analog high and low pass filters at cutoff frequencies of 0.1 Hz and 70 Hz, respectively. The recording system uses a 4 channel referential montage and used in continuous mode so that prolonged data sets containing ictal as well as interictal data could be collected for analysis. Five dynamical measures including nonlinear embedding delay τ, Hurst scaling α, wavelet scale ζ, autocorrelation measure ω, and gradient of accumulated energy ε were evaluated offline. The analyzed data comprised of 1-hr of EEG segment containing a minimum of one seizure and two hours of EEG segment from interictal period selected randomly.Results: The criteria chosen for the performance evaluation were, high statistical robustness as determined through the predictability index, the sensitivity and the specificity of a given measure to detect an EEG seizure, the lag in seizure detection with respect to the EEG seizure onset time, as determined through visual inspection and the computational efficiency for each detection measure. An optimality function was designed to evaluate the overall performance of each measure dependent on the criteria chosen. In the attached figure, we show the time trace of each dynamical measure evaluated on a sample EEG trace of 5 minutes in duration containing a seizure. Finally the table below summarizes the results of this study. Conclusions: The nonlinear embedding delay measure was found to have the highest optimality index due to its ability to detect seizure very close to the EEG seizure onset time, thereby making it a suitable measure for online seizure onset detection.