Abstract
Abstract: :
Purpose: Harwerth, et al. (IOVS 45:3152, 2004) used psychophysical and histological measurements in normal and glaucomatous monkey eyes to develop a mathematical relation between visual sensitivity and ganglion cell (GC) density. This study was designed to test the Harwerth formalism in patients with glaucoma by comparing visual threshold measured with achromatic perimetry and ganglion cell layer (GCL) thickness measured with optical coherence tomography (OCT). Methods: Patients with mild to moderate visual field damage were recruited for this study. Measurements were restricted to the central 10° where normal GCL is thickest. Visual sensitivity was measured with the 10–2 full threshold algorithm. Points with thresholds less than 10 dB were excluded. Compensation for the displacement of GCs from photoreceptors in the fovea (Sjostrand et al., Graefes Archiv 237:1014, 1999) was applied to the field points. StratusTM OCT images were obtained from a 6x6 mm square centered on the fovea using 10 evenly spaced horizontal b–scans of 512 a–scans each. The border between the GCL and IPL was indistinct, but the inner border of the GCL and the outer border of the inner plexiform layer (IPL) were detected with proprietary algorithms and a thickness equal to GCL+IPL calculated. Interpolation was used to estimate the GCL+IPL thickness at the field points. Predicted GC density was calculated from visual sensitivity using the Harwerth formalism with no change in parameters. To account for the IPL, a linear function of eccentricity was subtracted from the GCL+IPL thickness. The slope and intercept of this function were allowed to vary. A nonlinear least square procedure minimized the difference between the predicted GC density and measured thickness. A third adjustable parameter, which can be interpreted as ganglion cell spacing, was added to shift the two data sets relative to each other. Results: For eccentricities >4°, GC density predicted from visual sensitivity and GCL thickness calculated from OCT measurements followed the Harwerth formalism (r2 = 0.3 – 0.5). More centrally, where the Harwerth formalism represents an extrapolation, the measured thickness was always less than the predicted. Estimated GC spacing was about 10 µm, an anatomically reasonable value. Conclusions: The Harwerth formalism linking visual sensitivity to ganglion cell density describes the relation between paracentral glaucomatous visual fields and OCT–derived GCL thickness. Some alteration in Harwerth parameters or a more accurate IPL thickness may be necessary to extend the model centrally.
Keywords: visual fields • ganglion cells • imaging/image analysis: clinical