Martin T W Scott¹, Hui Xu¹, Alexandra Yakovleva¹, Robert Tibshirani¹, Jeffrey L Goldberg¹, Anthony M Norcia¹; ¹Stanford University

The ON and OFF retino-cortical pathways encode increments and decrements (respectively) in perceived luminance. Recent evidence from small animal models and human electrophysiology suggests that the OFF-pathway is more vulnerable to damage from Glaucoma than the ON-pathway. Thus, OFF-pathway based measurements of visual function may be useful in the diagnosis of Glaucoma. The steady-state visually evoked potential (SSVEP) measured with high density electroencephalography (EEG) can be used to non-invasively make these functional measurements. Here, in a cohort of 98 Glaucoma patients and 71 age-matched controls, we examine the extent to which SSVEP measurements obtained using ON- and OFF-pathway biasing visual stimuli are useful for case-control classification. Full-field SSVEPs were obtained using a spatial array of low-contrast, flicker-frequency tagged hexagonal probes. The luminance of these probes was modulated at 2.7Hz with a saw-tooth function, the fast phase of which biased SSVEPs towards either the ON- or OFF-pathways. Using both a logistic regression and a random forest classifier, we report moderate overall classification performance, with SSVEP features evoked by OFF-pathway biasing stimuli producing a small improvement in predictive accuracy over ON-pathway biased SSVEP features. Perhaps more importantly, classification based on signal phase features significantly outperformed classification based on signal amplitude features, particularly when OFF-biasing stimuli were used. While the classification performance we obtain is not yet sufficient for direct clinical use, it highlights the high information value of phase-related features, a metric that is often omitted in existing related work using the isolated-check VEP.

Acknowledgements: Research to Prevent Blindness, NIH (R01-EY030361-01, P30-EY026877)