Abstract
Purpose :
To examine the repeatability and reliability of the novel Foraging Interactive D-prime (FInD) test for the contrast sensitivity function (CSF) against a standard Two-Alternative Forced Choice (2AFC) method. Compared to other tests, FInD is self-administered, generalizable, and user-friendly for patients of all ages, cognitive status, language backgrounds, and literacy levels.
Methods :
FInD was used to present three charts of 4*4 cells, a random subset of which contain band-pass filtered noise targets (peak spatial frequency 0.5, 1, 2, 4, 8, and 16 cpd) that observers are required to mouse-click. The contrast of each target is selected from a range spanning difficult (d’=0.1) to easy (d’=4.5), that is adaptively updated after each chart based on a decision function fit to Hit/Miss/Correct_Rejection/False_Alarm scores for each cell. Fourteen adult observers with varying visual acuities performed the FInD CSF and 2AFC CSF in random order at 40 cm and at 4 m. The FInD CSF test was repeated twice at both viewing distances. Repeatability was assessed using mixed linear models and Bland-Altman analysis. Reliability was assessed using mixed linear models.
Results :
The FInD CSF test was completed significantly more quickly than the 2AFC (4.98 ±1.52 min vs 10.86 ± 0.91min at 40 cm; 4.93±2.00 min vs 12.69 ± 2.11min at 4 m). A mixed linear model repeated measures ANOVA with Huynh-Feldt correction showed good repeatability for the FInD CSF test at 4 m (F=0.044, p=0.877, η2=0.002) and 40 cm (F=0.633, p=0.498, η2=0.024). Posthoc analysis showed a significant difference in test-retest repeatability for 0.5 cpd only at 4 m. Bland-Altman analysis showed proportional biases for 0.5 cpd at 40 cm (p=0.003), and 16 cpd at 4 m and 40 cm (p=0.008, p=0.002). The mixed linear model showed no overall significant difference between the FInD CSF and 2AFC CSF tests at 4 m (F=0.913, p=0.366, η2=0.034) and 40 cm (F=2.130, p=0.123, η2=0.098). Posthoc analysis showed significant differences at 0.5 cpd (p<0.001), 1 cpd (p=0.001), and 2 cpd (p=0.025) at 4 m. At 40 cm, posthoc analysis showed differences for 4 cpd (p=0.032), and 8 cpd (p=0.002).
Conclusions :
Good repeatability and reliability were found for the FInD CSF test at 4 m and 40 cm viewing distances. Completion of data collection (aim is n=50) will conclude whether the FInD CSF test is indicated for use in clinical practice.
This is a 2021 ARVO Annual Meeting abstract.