Quantification of Cataract Using the Retroillumination Module of the Nidek EAS-1000 Anterior Segment Analysis System

Abstract

Cortical and posterior subcapsular (PSC) cataracts are best quantified by a retroillumination method. We have used the Nidek EAS-1000 camera system to document such cataracts and have investigated how variability from operator input may be minimised when analysing retroillumination photographs. We then used our method to quantify the reproducibility of (i) image analysis and (ii) the system as a whole. A total of 99 eyes from 70 patients with either cortical or posterior subcapsular cataracts were photographed twice at one session. The first set of images was analysed twice and the second set once only. Cataract was classified as a density at least 14 grey scale units (cct) greater than clear lens. The brightest 2% was excluded from analysis as it corresponded to the first Purkinje image of the fixation light. The reproducibility and ‘noise’ of the analysis method were assessed by calculating the correlation coefficient (r²) and standard deviation (SD) of the differences obtained for the two analyses of the first set of images. Replicate analysis from the digitised images was excellent (r²=0.99, SD of the opacities differences ± 1.74%). For the analysis of the system as a whole, replicate photographs showed a greater variability but still good: r²=0.94 and the SD of the differences ± 4.69%. The reproducibility of the system is limited by variability of the image capture rather than by the image analysis methods.

Measured Changes in Cataract over Six Months: Sensitivity of the NidekEAS-1000

“…Analysis was done in masked fashion with no knowledge o f which images were being analysed except where re-analysis was necessary for some retroillumi nation pictures because o f differing pupil diameters. Retroillumination images were analysed according to a protocol previously described [20]. If there was a dif ference between initial and 6-month pictures, the smaller of the two diameters was used for both pic tures.…”

Reproducibility of automatic standard digital analysis of lens opacities

“…Previous studies have shown the EAS-1000 t o be highly reproducible112 in lenses with varying degrees of opacity. 2 We have used custom-written standardized soft-ware for image analysis to minimize observer variation. For the longitudinal analyses t o be used by the VECAT Study it was necessary t o examine the repeatability of the measurements utilizing this software in lenses with little or no significant opacity.…”

“…Previous studies have also indicated the capture and analysis of Nidek EAS-1000 images to be reproducible. 1 2. Analysis of components of variance7 is presented inTable 1.…”