Abstract
Abstract: :
Purpose: In conventional perimetry, both differential light sensitivity (DLS) and stimulus steps are scaled in decibels (dB) and variability is highest in damaged regions of the visual field. When converted to linear sensitivity, DLS has a more linear relationship with structural measures of glaucomatous neuropathy and variability is highest in normal regions of the visual field. Computer simulations support the hypothesis that lower variability can be obtained by using linear stimulus steps. Methods: We tested one eye each, for 21 perimetrically–experienced patients with glaucoma of varying severity (MD range 0 to –23 dB, mean –7 dB) and 20 age–similar control subjects on 3 separate occasions. The first visit was used to train participants on our psychophysics apparatus. Test–retest variability was calculated from the latter 2 visits. DLS was measured at 8 locations (at 3 eccentricities, 9.5, 15 and 21 degrees) using a dB staircase strategy similar to the Humphrey Full Threshold (FT) strategy and our new Linear strategy (LnS). DLS was recorded in units of 1/Lambert (L–1). Our LnS terminated after 2 reversals and used a stimulus step of 500 L–1 before the first reversal and 250 L–1 thereafter. The mean sensitivity for the two tests was computed for each location. Overall test–retest variability was computed as the standard deviation of difference in sensitivities at these two visits for two sets of locations across all subjects: locations with sensitivity ≤500 L–1 (27dB) and those with sensitivity > 1000 L–1 (30dB). Locations where the maximum stimulus was not seen on both occasions were excluded from variability analysis. Results: Mean sensitivities for the control group ranged from 700–1200 L–1 for FT and from 700–1400 L–1 for LnS. Pointwise test–retest difference reached as high as 3600 L–1 for FT but only 1500 L–1 for LnS. For sensitivities ≤ 500 L–1 overall test–retest variability was similar for both strategies (F = 1.10, p = 0.56), but for sensitivities > 1000 overall variability was lower for LnS (F = 3.34, p < 0.001). There was a lower dependence of pointwise test–retest difference on mean sensitivity for LnS compared with FT. Across all subjects, mean test time was similar for FT and LnS. Conclusions: The use of perimetric strategies which employ linear stimulus steps can reduce extremes of variability in normal regions of the visual field, giving rise to relatively uniform variability characteristics across the sensitivity range and potentially enabling earlier detection of glaucomatous damage.