Abstract
Stereopsis is a critical visual function, however clinical stereotests are time-consuming, coarse in resolution, suffer poor test-retest repeatability, or agreement with other tests(Conner & Tidbury,2018). We developed AIM-Stereoacuity in which observers report via mouse the orientation of a 5x1.25𐩑 bar defined by the disparity of 100 elements embedded in a 6⌀𐩑 circle, and presented in a 4*4 grid of cells. The disparity of each bar is scaled from ±2σ relative to experimenter-set threshold-estimate in chart one and thereafter based on a fit to data from subsequent charts. Orientation errors(i.e. indicated vs. actual bar-orientation) as a function of horizontal disparity were fit with a cumulative Gaussian function to derive stereo-thresholds. 21 normally-sighted observers were tested with a red/blue anaglyph display in crossed and uncrossed disparity signs with 4 different element-types(8.5arcmin broadband dots, or band-pass difference of Gaussians with SFpeak@2,4,6cyl/𐩑). We analyzed stereoacuity-thresholds, durations, test-retest repeatability, and compared thresholds to Randot and Titmus ring-stimuli tests. Averaged across SFs and observers, mean-durations were 490.2sec and 533.0sec(mean chart-duration=42.6sec) for measuring crossed and uncrossed disparity, respectively. There was no effect of disparity sign or SFpeak(2-Way-ANOVA;p>0.05). Mean stereo-thresholds averaged across all SFpeaks were 92.7arcsec and 92.8arcsec for crossed and uncrossed disparities, respectively. Mean-thresholds using Randot and Titmus were 27.4arcsec and 41.4arcsec, respectively, and were not significantly correlated(r=0.07), nor with any element type(r=0.01-0.54; Mean=0.20). Crossed and uncrossed disparities were moderately correlated(r=0.10-0.83; Mean=0.56) with each other. Test-retest biases were -45.6arcsec and -49.3arcsec for crossed and uncrossed disparities, respectively. AIM-Stereoacuity showed reliable stereo-threshold results and is able to deliver adaptive, self-administered stereoacuity results in less than 2min.