Calculating the predicted retinal thickness from spectral domain and time domain optical coherence tomography - comparison of different methods

To compare the accuracy of different methods of calculating predicted central retinal thickness values in order to allow comparison between results of spectral-domain optical coherence tomography (SD-OCT) and time-domain OCT (TD-OCT) devices. In a prospective cohort study, 100 consecutive healthy individuals without ocular disease underwent sequential scanning with SD-OCT (Spectralis OCT) and TD-OCT (Stratus OCT). A group of 60 eyes was used to generate the conversion equations, which were tested on the remaining 140 eyes. Four equations were used: 1. Mean difference between SD-OCT and TD-OCT; 2. Multiplying a ratio by the original retinal thickness; 3. Linear regression analysis using retinal thickness; and 4. Regression analysis using retinal thickness and spherical equivalent. All four methods were used to calculate predicted SD-OCT values from TD-OCT measurements, and vice versa. For all four equations, the predicted SD-OCT central retinal thickness values were similar to the actual SD-OCT, with mean difference ranging from 0.78 to 1.01 mu m, and intraclass correlation coefficients > 0.88. Both regression equations and mean difference showed greater accuracy, with variation between calculated and actual retinal thickness values a parts per thousand currency sign5 mu m in 60 % of eyes. In contrast, the ratio method was less accurate, with 15.8 % of eyes showing differences > 15 mu m. Similar results were found for predicted TD-OCT values. Several methods can be used to convert central retinal thickness values from SD-OCT to the predicted TD-OCT value, or vice versa, with high degrees of accuracy and reliability. These methods may allow comparison of OCT values from SD-OCT and TD-OCT devices in clinical trials and standard patient care.