Statistical modelling of the central 10-degree visual field in short-wavelength automated perimetry.

Background: Reports of short-wavelength pathway dysfunction in retinal eye disease suggest that short-wavelength automated perimetry may be a useful technique for the investigation of central visual function. The aim of this study was to adapt existing statistical procedures used for the investigation of 30-2 short-wavelength automated perimetry to the 10-2 program of the Humphrey Field Analyser.

Methods: A four- or six-point linear interpolation procedure was used to calculate normal visual field sensitivity for each of the 68 stimulus locations of the 10-2 program using empirical normal data from 51 normal subjects examined using the 30-2 program. Prediction limits for normality were derived at each stimulus location, enabling the calculation of age-corrected global perimetric indices and construction of probability maps for diffuse and focal visual field loss. The normal database was validated by empirical data from five normal subjects, stratified for age.

Results: The pointwise distribution of normal sensitivity exhibited a Gaussian distribution at the majority of stimulus locations. The pointwise coefficient of variation did not vary significantly across the visual field. Examples of diabetic pseudophakic patients and a patient with age-related macular degeneration are presented to illustrate the effectiveness of SWAP at detecting visual field abnormality in the central visual field.

Conclusion: Ten-degree SWAP is able to effectively detect focal visual field loss in central retinal eye disease which may precede those found using conventional perimetry. SWAP may prove to be an invaluable technique for the investigation of central retinal eye disease.