Background: The present study aims to investigate an automated qualitative and quantitative assessment system (Automated Quantification of After-Cataract [AQUA II]) of posterior capsule opacification (PCO) in high-resolution digital retroillumination images and consequently reduce observer bias and increase accuracy of PCO grading. Methods: A data set of 100 eyes with no to severe PCO was analysed. Ten eyes were consecutively photographed twice and ten images were rotated to give a total of 120 images for PCO assessment. Validity was determined by including subjective grading and repeatability was determined by evaluating the 20 additional images. Evaluation of posterior capsular opacification (EPCO), posterior capsule opacity (POCO) and AQUA I methods were included for comparative analysis of the data. Results: The system developed proved to classify six types of PCO. Validity was confirmed by a Pearson correlation coefficient of r = 0.95 (EPCO r = 0.93; POCO r = 0.72 and AQUA I r = 0.94). Repeatability was better in AQUA II (95% confidence interval [CI] for mean difference: 0.5 ± 1.2) than in subjective grading (95% CI for mean difference: 0.6 ± 1.7), in EPCO grading (95% CI for mean difference:-0.2 ± 1.5), in POCO grading (95% CI for mean difference: 1.6 ± 2.7) and in AQUA I (95% CI for mean difference:-1.1 ± 1.9). Conclusions: AQUA II is a system that for the first time not only objectively quantifies PCO, but also qualitatively assesses PCO in an automated manner with texture classification. AQUA II showed an excellent validity and repeatability.
ASJC Scopus subject areas
Fields of Expertise
- Information, Communication & Computing