Purchase this article with an account.
Laura Ottobelli, Luca Mario Rossetti, Antonio Modarelli, Paolo Fogagnolo, David P Crabb, Giovanni Montesano; Structural mapping error in non-fundus perimetry. Invest. Ophthalmol. Vis. Sci. 2018;59(9):4061. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To use fundus perimetry to quantify mapping error of testing locations when precise localization from fundus tracking is missing.
31 glaucoma subjects with macular damage and 17 normals underwent dense macular OCT scan (Spectralis, Heidelberg Engineering) and a 10-2 visual field with a fundus perimeter (Compass, CenterVue), with the testing grid centred on preferred fixation. Fundus images from the two devices were matched to obtain precise localization of the tested locations on the OCT Ganglion Cell Layer (GCL) map (Figure 1 A - B). We accounted for Henle fibre displacement using the Drasdo model. The anatomical fovea, and not the centre of the grid, was used as the centre for displacement (Figure 1 D). We compared this Fundus Guided (FG) mapping with the usual blind mapping assuming a grid perfectly centred on the fovea and rotated along the fovea – disc axis (Figure 1 C).We measured the error as the distance between displaced stimuli locations with the blind mapping and the respective true locations from the FG mapping.The mapping error was analysed using a generalized linear mixed model at different eccentricities from fixation.
In normal and glaucoma subjects the error of blind mapping increased with eccentricity (Figure 2, p < 0.001). Due to eccentric fixation (Figure 1 D), error in glaucoma subjects was larger (p < 0.05 beyond 1.4 degrees). With no rotation along the fovea – disc axis, the error was significantly reduced (all p < 0.001, Figure 2), with the smallest improvement at 1.4 degrees (10% in glaucoma and 6% in normal subjects) and the largest at 9.1 degrees (71% in glaucoma and 65% in normal subjects).
Grid rotation along the fovea – disc axis is an unreliable assumption. When central damage is present, glaucoma subjects exhibit eccentric fixation, making blind mapping inaccurate.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.
Patient example: A) Stimuli location on the Compass fundus image; B) Spectralis matched fundus image overlaid; Drasdo displacement with Blind mapping (C) and Fundus guided mapping (D). The colour map represents GCL thickness. Real stimuli locations are shifted toward the residual ganglion cells due to eccentric fixation (D).
Mapping error at different eccentricities. Peripheral error is higher with rotation along the fovea – disc axis (left panels) compared to non rotated grids (right panels). Boxes enclose the interquartile range, whiskers extend to the 5th and 95th percentiles.
This PDF is available to Subscribers Only