Average lens density quantification with swept-source optical coherence tomography: optimized, automated cataract grading technique

Purpose: To assess a revised version of an objective cataract grading method based on the lens densitometry on sweptsource optical coherence tomography (SS-OCT) scans. Setting: Rothschild Foundation, Paris, France. Design: Prospective case series. Methods: All patients seeking a cataract evaluation and consenting to study participation were included. Exclusion criteria were a history of ocular surgery, cornea or retina disorders, and ocular dryness. The average lens densitometry (ALD) was measured on SS-OCT scans (IOLMaster 700) using an algorithm developed with MATLAB software. The ocular scatter index (OSI), measured using a double-pass aberrometer (Optical Quality Analysis System); Pentacam nucleus staging (PNS) score; and mean PNS were also measured and compared with the ALD measurements. Results: Two hundred eighty-five eyes (195 with cataract; 90 controls) of 155 patients aged 20 to 93 years (mean 56.5 years ± 21 [SD]) were included. The ALD measurements were repeatable (P = .99, analysis of variance). The repeatability limit was 2.50 pixel units. The ALD correlated with the OSI (r2 = 0.41, P < .01) and PNS score (r2 = 0.65, P < .01). An ALD greater than 73.8 pixel units was the cutoff threshold for cataract, with 96.2% sensitivity and 91.3% specificity. Discriminant analysis produced a model combining visual acuity and the mean ALD with an area under the receiver operating characteristic curve of 0.975 (P < .001), where the mean ALD had the highest coefficient. Conclusions: Using the automated algorithm based on SS-OCT ALD measurements was a repeatable, reliable objective cataract grading method. When the ALD is greater than 73.8 pixel units, it is reasonable to discuss cataract surgery in symptomatic patients.