A Prospective Validation Study of an Automated Ocular Proptosis Device in Thyroid Eye Disease

BACKGROUND: Accurate measurement of proptosis is crucial for diagnosing and managing thyroid eye disease (TED). While the Hertel exophthalmometer remains widely used, it demonstrates substantial interobserver variability. This prospective diagnostic accuracy study aimed to validate a novel automated ocular proptosis device against computed tomography (CT) as the reference standard. METHODS: We enrolled 102 patients (204 eyes) with TED who underwent proptosis measurement using 2 methods: orbital CT, an automated device, senior ophthalmologists (>5 y of experience), and junior ophthalmologists (1-3 y of experience). Primary outcomes included intraclass correlation coefficient (ICC), Bland-Altman analysis with 95% limits of agreement (LOA), mean absolute error (MAE), and root mean square error (RMSE). RESULTS: The automated device demonstrated excellent agreement with CT [intraclass correlation coefficient (ICC) = 0.985, 95% CI: 0.980-0.989], superior to both senior (ICC = 0.934) and junior (ICC = 0.912) physicians. Mean bias (device versus CT) was minimal at -0.08 mm with 95% limits of agreement (LOA) of -1.65 to +1.49 mm. In comparison, senior physicians showed significant positive bias (+0.72 mm), while junior physicians exhibited wider LOA (-3.15 to +2.53 mm). The device achieved the lowest MAE (0.61 mm) and RMSE (0.78 mm), indicating superior accuracy. Interobserver reliability was significantly higher for the automated device (ICC = 0.992) than for manual measurements. CONCLUSIONS: These findings demonstrate that the automated ocular proptosis device provides highly accurate and reproducible measurements, outperforming manual assessments by ophthalmologists across experience levels. Its excellent agreement with CT supports its potential as a new standard for objective proptosis assessment in TED clinical practice and research.