Abstract
Purpose :
Simulation is the cornerstone of assessing new perimetric algorithms, enabling exhaustive testing of the reliability and accuracy of new procedures. We developed a novel simulation algorithm, the Kinetic Adaptive Numerous Gamma Algorithm (KANGA), which simulates kinetic perimetry responses based on an individual’s static perimetry thresholds.
Methods :
Using the Open Perimetry Interface on the Octopus 900 perimeter, we measured size III white-on-white static thresholds at 10 degree spacing for 3 individuals with normal vision (34-46 years old) along 6 kinetic vectors. Thresholds were taken as the average of 5 ZEST procedures. We fitted these with a parabola. We also measured kinetic responses to size III 27dB stimuli, moving from peripheral (unseen) to central (seen) locations at 3 deg/s. A low 27dB luminance level was required to ensure that kinetic thresholds fell within the measured static field of 60 degrees eccentricity. We fit a sum of gamma distributions to the distance between the 27dB static threshold location and the kinetic button press (corrected for reaction time). Simulated kinetic responses were based on an interpolated static threshold location plus a random sample from that gamma distribution. After fitting the model, we validated the predictive power of KANGA by comparing the mean and standard deviation between virtual and human kinetic responses (10 observers aged 21-46) to size III stimuli over a range of luminance values (20, 24, 25 and 30dB). KANGA was seeded with the average static fields of the same observers and generated 1000 simulations per vector.
Results :
The model of best fit produced a mean difference of 0.66 degrees across 6 vectors (90 kinetic button presses to size III 27dB stimuli), fitted to the first 3 observers (with a standard deviation of 6.64 degrees). Taking these model parameters to simulate responses for 10 observers over a larger range of target luminance values (20, 24, 25 and 30dB), KANGA produced an error between -2.4 and -9.1 degrees (depending on the luminance value of the target), and a standard deviation ranging between 3.4 and 6 degrees.
Conclusions :
KANGA provides a platform for simulating observer responses to kinetic stimuli and may lead to further development of efficient procedures for assessing visual field loss in both at risk and diseased eyes.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.