Threshold versus intensity functions in two-colour automated perimetry.

Abstract

Two-colour computerised perimetry is a technique developed for assessing cone- and rod-function at fixed background luminances in retinal disease. However, the state of adaptation during testing is unknown but crucial in the interpretation of results. We therefore aimed to determine the adaptational state of rod- and cone-mechanisms in two-colour perimetry. Sensitivity to 480nm (blue) and 640nm (red) Goldmann size V targets was determined for 10 normal subjects aged 16 to 46years at 17 locations in the central 60 degrees of the visual field under scotopic conditions and then from -1.5 log cd m-2 to 2 log cd m-2 (white background) in 0.5 log unit steps. Data were fitted with threshold versus intensity (tvi) functions of the form logT=logT0 + log ((A+A0 )/A0 )n . No clear rod-cone break was observed for 640nm stimuli. For 480nm stimuli, transition from rod-detection to cone-detection occurred at mesopic illumination levels, where rod adaptation approached Weber behaviour. Cone detection mechanisms did not display Weber-like adaptation until the background luminance approached 1 log cd.m-2 . Diseases resulting in a "filter effect" - including disorders of the photoreceptors - are therefore predicted to affect sensitivity when rod function is probed with short-wavelength targets under scotopic conditions, but less so under mesopic conditions. Filter effects are similarly anticipated to affect cone function measured using long-wavelength targets under mesopic conditions (e.g., during microperimetry), but less so under photopic conditions. Asymmetries in adaptation in automated two-colour perimetry are predicted to artefactually favour the detection of losses in rod sensitivity under scotopic conditions and cones under mesopic conditions.