Reproducibility of an Analysis Software for Qualitative Observation of Elschnig Pearls

BACKGROUND AND OBJECTIVE: To determine the reproducibility of a semi-objective analysis software for the morphological analysis of single Elschnig pearls from retroillumination images of posterior capsule opacification. PATIENTS AND METHODS: In this prospective study, images of eyes with varying degrees of regeneratory posterior capsule opacification and the presence of Elschnig pearls were taken. The size and morphology of single Elschnig pearls was evaluated from consecutive images using an analysis method called PearlTracer using three different modes: snake selection, where a few points detected the pearl border automatically; diameter circle selection, a circle function; and polygon selection, where several points were connected by straight lines. The interobserver and intraobserver reproducibility and the limits of the different modes concerning contrast were assessed. RESULTS: One hundred fifty-eight retroillumination images of 55 eyes of 50 patients were included with a total of 857 pearls being analyzed. A high interobserver reproducibility was found, as well as a high intraobserver reproducibility. Furthermore, a significant difference in the pearl size and the pearl morphology between the three analysis features of the software was found. CONCLUSION: For the observation of single Elschnig pearls and their morphological changes, the semi-objective PearlTracer analysis software was found to be a highly reproducible and useful method.