Comparison of linear, multilinear and mask microdensitometric analyses of Scheimpflug images of the lens nucleus.

We wanted to determine the agreement among three methods currently used to measure the average density of the nucleus from a Scheimpflug video image of the lens, namely, the multilinear, linear and mask techniques. Using the NEI Scheimpflug Cataract Imaging System, video images of the lens were obtained from 49 eyes of 31 consecutive patients with predominantly nuclear opacities of varying severities. For each image, 3 separate determinations of the average nuclear density [in optical density units (odu)] were obtained by one analyst sequentially applying each of the three methods. Repeat measurements were obtained by the same analyst one week apart to evaluate reproducibility (using 95% error range), of each of the methods. Pairwise comparisons of the 3 methods were also done. Only the error due to outlining of the nucleus was assessed. Reproducibility of each method was excellent. The 95% error ranges were +/- 0.015 odu for the linear, +/- 0.013 odu for the multilinear, and +/- 0.017 odu for the mask methods. Very strong linear relationships were observed between the linear and multilinear methods (R2 = 0.99), between the linear and mask methods (R2 = 0.97), and between the multilinear and mask methods (R2 = 0.98). Simple linear regressions calibrating these three methods against each other would facilitate comparison of density measurements using any of these methods for analysis of Scheimpflug images of the lens nucleus.