Mathematical approximation of Orbscan II central corneal thickness to contact ultrasound

To compare Orbscan II central corneal thickness (CCT) measurements corrected by five mathematical methods with ultrasound (US) CCT measurements.

The preoperative CCT examinations of the right eyes of 3242 consecutive candidates for refractive surgery were retrospectively reviewed. Raw Orbscan data were processed by five correction methods: multiplication, subtraction, polynomial, logarithmic, and power. For each correction method, the concordance between corrected Orbscan and US pachymetries was assessed by mean, Pearson correlation, regression analysis, 95% limits of agreement (LoA), and percentage of eyes within 8 μm of US pachymetry.

Throughout the 3242 eyes analyzed, we found a high Pearson correlation between Orbscan and US measurements with all correction methods assessed. The linear multiplication and subtraction correction methods brought about mean Orbscan CCT similar to mean US CCT measurements. However, they led to relatively high 95% LoA intervals and low percentage of eyes within 8 μm of US pachymetry. In contrast, the three nonlinear correction methods proposed herein for the first time narrowed the 95% LoA interval (Bland-Altman analysis) together with providing mean Orbscan CCT similar to mean US CCT measurements. Moreover, the power fit was the only method that was reproducible when extracted from one subset of the cohort and applied in another.

Broad 95% LoA intervals were obtained when applying the multiplication and the subtraction correction methods, indicating a poor agreement between the devices. We suggest the power fit as the default method for adjusting Orbscan to US CCT readings.