Automatic detection of ocular movements during sleep with an application to a sleep-disorders study

An algorithm for automatic detection of ocular movements (OMs) during sleep is described. Typical EOG recordings were transformed into the time-frequency plane using a wavelet transform. This kind of representation was helpful to localize preciselythe OMs activity in the frequency domain, making thedesign of two digital filters (FEOG1 and FEOG2) possible. The EOG signal was filtered off-line using FEOG1 and FEOG2. Then 2 context-dependent thresholds were computed to perform OMs detection. Chin EMG recordings were processed simultaneously in order to reject movements due to artefacts. The automatic detection was applied to a pharmacological study aimed at measuring the action of 2 pharmacological compounds on sleep disorders.