Abstracts

Rationale: Ictal surface electroencephalogram(EEG) is a useful investigation in the presurgical evaluation of refractory epilepsy patients. The presence of (1)artefacts and (2)the wide ranging patterns of ictal discharges are two hurdles to producing good diagnostic accuracry. Quantitative EEG may aid electrographers in localizing and lateralizing seizure onset. This study used the (1)wavelet analysis and (2)absolute slope method to overcome these hurdles. Methods: We retrospectively selected 12 patients who achieved good surgical outcome following temporal lobe surgery at 1 year. Each patient had between 1-3 ictal episodes analyzed that were their habitual seizures. The EEG signals were given decimated discrete wavelet transform(DWT) using biorthogonal wavelet 3.9. The approximation coefficients at level 4 were obtained and the EEG signals were reconstructed using inverse wavelet transform to produce an artefact removal effect. EEG signals were arranged in bipolar montage and the absolute slope was calculated between signals which ran over all channels. They were then smoothed by a lagging moving average of 5s duration. This property of the EEG signals increases for both high-amplitude slow and slow-amplitude fast activities. The absolute slopes were then normalised according to the means and standard deviations obtained from a 30s reference period. Ictal discharges were defined as S(t)>2.5. The earliest channel with ictal discharges was selected if at least one adjacent bipolar channel was involved and the duration of ictal discharges was >20% of the seizure time. For the purpose of localization, any one of the constituent channels of a group of channels belonging to a specific region to display this property first will be selected as ictal onset region (temporal, frontal, parietal, occipital). Two board certified neurologists independently inspected the seizure episodes blinded to clinical information and made decisions about the lateralization/localization information using bipolar montage, paper speed 30mm/s, gain 100uV and band/notch filters. Results: 25 seizures were analyzed. Using resection results as standard, 16 seizures(64%) were correctly localized by electrographers using visual analysis. By comparison 22 seizures(88%) were correctly localized by quantitative EEG. Using resection results as standard, 15 seizures(60%) were correctly lateralized by electrographers using visual analysis. By comparison 24 seizures(96%) were correctly laterlized by quantitive EEG. Overall, 14 seizures(56%) were both localized and lateralized correctly by electrographers whereas 21 seizures(84%) were both localized and lateralized by quantitative EEG. Conclusions: Coupling DWT with the absolute slope method helps clinicians achieve a better diagnostic accuracy with ictal surface EEG.