Towards StabilEyes: Detecting Repetitive Pupil Motion in Nystagmus using Discrete Period Quadrature

Abstract

We present a novel method to determine frequency and phase of repetitive eye motion, using the widely available platforms of smartphones and tablets, requiring only the download of a free software app. Our purpose is to provide assistance to patients with nystagmus by presenting an image on the smart device’s screen that moves in synchrony with the patient’s periodic eye motion, thereby stabilizing it. The image may originate in real time from the device’s back-facing camera, thus providing a stable view of the visual environment. We have named the prospective app, "StabilEyes." The app would also gather eye motion data during activities of daily living to guide treatment and inform research into nystagmus and other diseases of the visual system. As opposed to most applications for tracking eye motion, ours is not based on first de-termining pupil location. Rather, our method simply identifies the central region of the face containing the eyes and then finds periodic variations in the first moment of pixel intensity within this region generated by movement of the iris and pupil relative to the visible portion of the sclera. We determine the frequency and phase of these variations using a novel method of time-frequency analysis, discrete period quadrature (DPQ), which combines aspects of the phase-locked loop and the discrete Fourier transform, by implement-ing a bank of phase-locked loops to yield a spectrum. Results are shown of applying the system to video image sequences that simulate the repetitive eye motions of nystagmus, reconstructed from still images of normal subjects gazing in predetermined directions.