Purchase this article with an account.
William H Swanson, Andrew J Anderson; Comparing Defect Depths for Different Perimetric Stimuli. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):1049. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Perimetric defect depths for sinusoids (e.g. frequency-doubling stimuli) must be converted into Weber contrast before being compared to defects for luminance increments (e.g. Goldmann stimuli). Shapley & Enroth-Cugell (1984) used (Peak-Trough)/Trough for this conversion, while King-Smith & Kulikowski (1975) used (Peak-Mean)/Mean. Here we compare the performance of these conversion methods.
One eye each was tested for 41 patients with glaucoma and 19 age-similar controls free of ocular disorders, on custom testing stations. The first dataset tested 11 patients using static 200 msec presentations of Gaussian blobs with standard deviations of 0.25° and 0.50°, and a Gabor sinusoid in cosine phase at 0.5 cycle/degree with a standard deviation of 0.50°. The maximum for (Peak-Mean)/Mean was 800% for the blobs and 434% for the Gabor. The three stimuli were compared at 30 locations within ±9° of the horizontal midline. The second dataset tested 21 patients using stimuli scaled with visual field location. Each Gaussian blob had a standard deviation that was ½ the standard deviation of the Gabor at that visual field location, and the two stimuli were compared at 14 locations across the visual field. The blobs were presented using static 200 msec presentations and the Gabors were presented as 3 cycles of 5 Hz counterphase flicker. The maximum for (Peak-Mean)/Mean was 800% for the blobs and 70% for the Gabors. The third dataset tested 20 patients with the same stimuli and locations as the second dataset, but with more stimulus presentations per location.
The Akaike Information Criterion favored the method of King-Smith & Kulikowski for all three datasets: log likelihoods (base 10) of 17.8, 2.1 and 2.5 for datasets 1, 2 & 3, respectively. Bland-Altman analysis of the first dataset found that disagreement became greater at higher threshold elevations for the method of Shapley & Enroth-Cugell (t = -4.92, p < 0.0001) but not for the method of King-Smith & Kulikowski (t = +0.67, p = 0.25).
These data provide support for comparing glaucomatous defects with sinusoids and luminance increments using the method of King-Smith & Kulikowski rather than the method of Shapley & Enroth-Cugell.
This PDF is available to Subscribers Only