Todd Hudson^1,2, John-Ross Rizzo^1,2, Janet Rucker¹; ¹NYU Grossman School of Medicine, ²NYU Tandon School of Engineering (BME)

Continuous smooth pursuit, implemented via sinusoidal visual target motion, is a task well-known in both neuroscience and clinical neurology. Typical measures extracted from smooth pursuit include velocity gain, and the presence/absence of saccadic pursuit. Here, pursuit was performed in two 12s epochs, starting at screen center and moving sinusoidally first in the righthand and then in the lefthand side of the screen (3 cycles each). We performed both a conventional and Bayesian spectral analysis of the pursuit response. A conventional measurement of smooth pursuit gain (velocity gain) is carried out by computing the sample-by-sample ratio of estimated eye speed along the axis of target motion, to the speed of the pursuit target. In contrast, Bayesian spectral analysis takes advantage of the built-in structure of the data, and measures the amplitude gain of the sinusoidal component of the response at the fundamental (.25 Hz) frequency. Further, we demonstrate the possibility of measuring response components that might occur at the harmonics, and along dimensions other than the (horizontal) axis of pursuit. Pursuit gain, as obtained in one example participant under these conditions, is measured with two orders of magnitude greater precision using the Bayesian relative to the conventional method. We can also see that the conventional method substantially underestimates pursuit gain [0.9 vs. 1.0] in our example. This method in addition allows us to subtract out the measured pursuit response, greatly simplifying the task of saccade detection. In our example data, we find microsaccades throughout the two pursuit epochs, but no saccades large enough to be reliably detected in a bedside neurological exam. This approach offers new opportunities for basic scientific understanding of the pursuit system during sustained tracking, and clinical discovery in, for example, Parkinson's and MS research, as these additional metrics and sensitivity relate to disease progression and prodromal detection.