EOG recordings of saccades

Ocular motion sensing using electrooculography (EOG) has been widely employed to measure different eye movements, fordiagnosing medical disorders such as sleep disorders and different psychological conditions. Electrooculography (EOG) is a body-wornsensor that has several unique capabilities, when compared with external sensors such as camera-based systems. Primarily, EOG canbe used to sense precise eye movements in dark conditions and closed-eye conditions. However, EOG signals are heavily contaminatedwith electroencephalography (brain signals), electromyography (extraocular muscle and face muscles), eyelid movements, skin drift,skin resistance, and other noise and artifacts. Traditional filtering approaches are unable to significantly improve EOG signals dueto the frequency overlap between information (e.g., saccades, smooth pursuit movements), and noise and artifacts. The present paperdevelops a methodology using a constant velocity model (where, the rate of change of the corneal-retinal potential of a saccadeis considered to be a constant) and an acceleration model (where, the double rate of change of the corneal-retinal potential of asaccade is considered to be a constant) to fit the corneal-retinal potential during saccadic eye movements. A Kalman filter (KF) isused for model-based sensor fusion. It is shown that the proposed methods are able to improve the signal-to-noise ratio (SNR) ofthe measured saccades of EOG signals by about 30-40% when compared to what is possible with finite impulse response filters andBrownian model-based KF methods.