Abstracts

RATIONALE: It has been reported that spikes during REM sleep (REM) have higher localizing value than spikes during other stages of the sleep-wake cycle. REM identification with conventional sleep staging methods is time consuming. We present a new method for rapidly identifying REM using EEG power spectrum. METHODS: Sixteen consecutive patients admitted to a tertiary care epilepsy center were studied. Along with standard EEG telemetry, each patient also had two nights of telemetry including chin EMG and EOG to allow sleep staging. Continuous power spectrum of EMG and EEG channels were generated using a 2.56s epoch length with a time resolution of 60 seconds. Technicians were instructed to select expected REM epochs using the following criteria: low delta, low spindle and low EMG band power. An EEGer then reviewed the EEGs to validate the recognition of REM epochs. REM epochs were assessed for the presence of spikes and seizures and the location of these was then compared to the overall assessment based on the clinical telemetry investigation. RESULTS: The method was specific in detecting REM epochs (83% of selected segments contained REM). The mean amount of REM detected was 69 minutes per night. Nine of 16 patients had no interictal epileptic discharges during their investigation. Of the seven patients with spikes, only two were observed to have spikes during REM. The spikes during REM in these patients were consistent with the results of the entire telemetry investigation. No seizures were recorded during REM. CONCLUSIONS: This simple method of detecting REM allowed the accurate identification of almost 70 minutes of REM per night. However, the frequency of spikes during REM in this group of patients was low. As well, spikes during REM did not demonstrate an improvement in localization over routine telemetry recordings.