Abstracts

Rationale: Evaluating the efficacy of absence seizure treatments has traditionally focused on comparing the frequency of seizures during treatment to seizure frequency during a finite baseline period. Electroencephalogram (EEG) recordings are used to supplement clinical observations of care givers. There is currently no reliable tool for rapid absence seizure counting which can quickly detect the absence seizure configuration in the current clinical environment. Furthermore, merely counting the number of seizures as a measure of treatment efficacy for absence seizures (defined for this purpose as spike and wave discharges > 3 seconds in duration) may not provide a full explanation of the therapeutic effect. Methods: In this study we propose an automated spike and wave discharge (SWD) detection algorithm based on time frequency analysis (Continuous Wavelet Transform) and variance statistics. Ambulatory EEG recordings in this study were acquired from two children <13 years of age; one seizure free (24 hours) and one experiencing seizures (4.5hours). Subjects were instructed to go about their normal life as usual while EEG recording was ongoing avoiding any type of activity that might result in loosening or removal of electrodes from the scalp or result in excessive recording artifacts, e.g., gum chewing. Results: SWD detected by the algorithm were compared to the SWDs manually scored by a board certified electroencephalographer. For the first patient we had only one false positive epoch during a 24h continuous EEG recording. For the second patient, our program detected 120 of the 150 manually scored 3 Hz SWD epochs. Twenty seven of the 30 missed epochs were <2.1 sec. The remaining epochs were 3.1, 3.3, and 4.1 seconds, respectively. All 3 missed epochs longer than 3 seconds turned out to be fragmented 3 Hz SWD with interruptions of <1 sec. Our algorithm had an error of <3% for detecting SWD >2sec, <2% for epochs >3sec and 0% for SWD epochs >5 seconds. Conclusions: Since clinically significant absence seizures are associated with SWD >3sec long this algorithm would be an attractive method for automatic detection and quantification of absence seizures. Applying this method in clinical trials of absence seizures may provide a reliable and accurate outcome measure of therapeutic interventions using minimal manpower. (Supported by: North Florida Foundation for Research and Education ,Inc. National Science Foundation and Air Force).