Removing Subjective Post-Mortem Grading from Posterior Capsular Opacification: A New Automated Detector Opacification Software, ADOS

Abstract

ABSTRACTPurpose/aim: Posterior capsular opacification (PCO) is the leading complication following cataract surgery. However, PCO grading methods vary between studies and rely on subjective scoring methods; hence, they are prone to error. Therefore, software that reduces subjectivity and enhances reproducibility is needed. The aim of the study was to evaluate the reproducibility and objectivity of a novel automated custom-designed PCO detection software.Materials and Methods: In total, 165 fixed eyes with intraocular lenses were sectioned to obtain a Miyake Apple view (MAV) image. The capsular bag (CB) with the in situ IOL was removed and imaged using the Olympus DSX110 stereoscope. Central, peripheral, and Soemmering’s ring area (SRA) and intensity (SRI) were graded by two ophthalmologists using a scale of 0–4. Software was developed to detect PCO, and these values were correlated with subjective scoring.Results: Intra-observer agreement for MAV and CB images ranged from poor to moderate, with very good agreement for SRI (k = 0.88, 0.86, respectively). Agreement between graders was fair to good (k = 0.31–0.77). Miyake Apple view and capsular bag images showed good correlation; central PCO showed the least correlation for both evaluators (k = 0.29, 0.72). Subjective average grading versus software correlation demonstrated very good correlation for intensity and good for area (r = 0.85, 0.61). Reproducibility of the methodology resulted in good to very good correlation.Conclusions: Software correlates with previous scoring methods and is a reliable and reproducible system. Moreover, the capsular bag view, as opposed to the Miyake Apple view, allows visualization of the capsular bag directly and avoids overlapping structures (vitreous, cornea) that may interfere with proper PCO quantification.