Can an LCD screen be a viable alternative to a projection bowl for clinical perimetry?

Abstract

Purpose : To assess the agreement between threshold sensitivities obtained from an LCD screen and a standard bowl perimeter

Methods : Achromatic contrast thresholds were measured using the method of constant stimuli (MOCS) at two visual field locations on a standard bowl perimeter (Octopus 900, Haag-Streit, UK, stimuli generated using R v3.2.2 and the Open Perimetry Interface v2.1) and a gamma-corrected 21.3” monochromatic LCD monitor (EIZO Radiforce GS521, refresh rate 51 Hz, stimuli generated using MATLAB [The Mathworks Inc, USA] and Psychtoolbox v.3.0) in 52 healthy and glaucomatous participants (MD range -11.60dB to +1.50dB). Goldmann III (SAP) stimulus contrast was varied at two locations of the VF: one central, one peripheral. Method of constant stimuli (MOCS) frequency-of-seeing (FOS) and a maximum likelihood method established threshold sensitivity (intensity seen 50% of the time). Mean difference between platforms and 95% limits of agreement (LoA) were calculated using Bland-Altman analysis and compared with test and retest differences for each platform and the 95% coefficient of repeatability (CoR). A threshold was generated from the first- and second-half of the data collected; test retest differences were calculated from the threshold estimates from each half.

Results : Threshold sensitivities were higher on the LCD monitor (mean difference 2.48 ± 3.40dB) with 95% LoA range being 13.50dB (Fig. 1). The 95% CoR was 17.42dB and 17.55dB on the LCD and bowl respectively. Throughout the sensitivity range (≥12dB), the difference between platforms was well represented (R²=0.81; p<0.001) by the orthogonal least-squares equation: 0.69xth_LCD - 0.72xth_bowl + 0.07xdist - 2.54=0; where, th=threshold and dist=vectorised distance from the point being tested to fixation. When queried, 12% of the patients reported to prefer the bowl surface, 21% had no preference, and 67% preferred to be tested on the LCD screen

Conclusions : Threshold measures captured using a bowl perimeter and LCD monitor while systematically different, can be equated following post-hoc adjustment. The wide limits of agreement observed between LCD and bowl perimetry was also explained by the poor repeatability for each platform independently

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

 

Figure 1. Bland-Altman analysis of the agreement between contrast thresholds measured using the LCD screen and the bowl surface