A new perimetry thresholding algorithm with size-modulated stimuli reduces variability by half in damaged regions of the visual field compared to SITA Standard

Abstract

Purpose : Identifying glaucomatous progression is challenging because of considerable variability in visual field (VF) test results. Reducing variability would reduce the time and/or number of VFs required to measure progression, improving clinical care and saving resources. Hypothesis: the precision (test retest variability) of a new thresholding algorithm (trail traced threshold test (T4; Zhu 2018; 59(9): ARVO E-abstract 5114 #5114) is better than that of the Humphrey Field Analyzer SITA Standard test (Carl Zeiss Meditec) for a similar test time. Interim results were presented Petriti 2018; 59(9): ARVO E-abstract 5115.

Methods : The T4 algorithm models a probability of seeing (PoS) curve at each location with a reversed cumulative normal distribution. The likelihood of response is a weighted binomial distribution in which the weights of neighbour locations are defined according to anatomical proximity; each response to a stimulus updates the PoS parameters at its own location and those of neighbouring locations. The algorithm was implemented on an EIZO GS521 monochromatic monitor. Stimuli were i) circular spot of light modulated in size at constant contrast (27dB) (T4sm) and ii) oscillating bars (T4mmdt). 86 clinically stable glaucoma patients with previous reliable VFs (median [IQR] age 69 [62-73] years and mean deviation -4.54 [-0.90 – -11.26] dB) underwent testing in one eye with SITA Standard, T4sm and T4mmdt in a randomized order; tests were repeated on the same day.

Results : SITA and T4sm are presented. Mean (SD) test time: SITA 5.51 (1.03) mins, T4sm 5.17 (0.52) mins. Figure 1: SD of test retest differences against location sensitivity. Variability (absolute test retest differences) is similar above 26dB threshold sensitivity (SITA 1.48dB (SD 0.22), T4sm 1.33dB (SD 0.56); p = 0.02) and reduced by half below 26dB (SITA 5.77dB (SD 2.53), T4sm 2.89dB (SD 1.02); p< 0.001). Figure 2: Bland-Altman plots of reproducibility (all locations, all subjects), left SITA Standard, right T4sm. The heteroscedasticity of test retest differences is considerably reduced.

Conclusions : The precision of T4sm is markedly better than SITA Standard in damaged locations of the VF for a similar test time. Improved precision and reduced heteroscedasticity should make a meaningful clinical impact, improving the detection and quantification of glaucoma progression.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.