Purpose
NETRA-G - a smart-phone based technology has been developed recently for quantifying the eye’s sphero-cylindrical refractive error. Refractive error is calculated from the alignment pattern of a pair of red-green lines made by the subject for 18 different axes between 0 - 180°. This study evaluated the minimum number and combination of axes required to determine refractive error, relative to the 18 axes estimates and gold-standard subjective refraction.
Methods
Sphero-cylindrical refractive error of 26 eyes of 15 volunteers (18-30yrs of age) was determined using subjective refraction and Netra-G by two independent, masked observers. Netra-G measurements were made on a Sony Xperia Ion® smart phone. The 18 axes data were then resampled using four strategies that involved progressively incrementing the number of axis I) in the order tested, II) by choosing only odd axes, III) by choosing only even axes and IV) by choosing axes from four distinct quadrants spanning the entire 180°. The experiment was repeated in 34 eyes of 17 volunteers (18-30yrs of age) who had little or no exposure to smart-phone technology using only the critical axes combinations determined earlier.
Results
The power vector values of refractive error, as determined by subjective refraction, ranged from -11.75D to 0.13D, -2.35D to 0.54D and -1.72D to 1.12D for M, J0 and J45 terms across all volunteers. For all four strategies, the difference in refraction estimates from the 18 axes measurements increased with a reduction in the number of axes used. Strategy IV performed closest to the 18 axes measurements. A minimum of 5 axes was required for Netra-G measurements to be within ±0.5D and 5° of gold standard values for M, J0 and J45 terms. The mean (±1SD) absolute difference between subjective refraction and Netra-G measurements for the M, J0 and J45 terms were -0.22±0.59D, 0.07±0.35D and -0.01±0.42D for the first dataset and 0.30±0.47D, 0.04±0.64D and -0.06±0.52D for the second dataset, respectively (p>0.03 for all).
Conclusions
Sphero-cylindrical refractive errors determined using Netra-G are within ±0.5D of the gold-standard subjective refraction. A minimum of 5 axes chosen using a quadrant-based strategy appears necessary for achieving this level of accuracy. Familiarity of subjects with smart-phone technology appears to have minimal influence on refractive error estimates obtained using NETRA-G.