Axial Length Estimation Using Metadata From Swept-Source Optical Coherence Tomography Angiography

Purpose: The purpose of this study was to explore the possibility of using metadata from optical coherence tomography angiography (OCTA) for axial length (AL) estimation. Methods: This retrospective, cross-sectional, observational study included patients with both OCTA imaging and biometry measurements available in a single tertiary medical center from 2019 to 2024. Univariate and multivariate regression analyses were used to build the estimation model of AL using metadata from OCTA. Results: A total of 198 eyes from 130 patients were included. The average AL was 24.58 mm (interquartile range [IQR] = 22.35-26.81). Univariate regression analysis identified that y_chinrest, z_chinrest, z_lens, and z_motor were all closely related to AL (P < 0.001). The final estimation model derived from multivariate regression analysis was Axial_Length = 24.79611 - 0.25510 × y_chinrest - 0.23443 × z_chinrest - 1.41824 × z_lens + 0.71947 × z_motor. The multiple R-squared value was 0.5586 and adjusted R-squared value was 0.5495. Of the predicted values, 68.7% were within 1 mm of the true AL, and 92.9% were within 2 mm of the true AL. Regression diagnostic plots confirmed assumptions of linearity, homoscedasticity, normality, and no multicollinearity were met. Foveal avascular zone (FAZ) correction with estimated AL showed a reduction in mean absolute error by 31%, from 0.031 mm2 to 0.021 mm2 (P < 0.001). Conclusions: The present study demonstrates the possibility of reliable AL estimation using metadata from OCTA.