Evaluation of a low-cost wavefront aberrometer for measuring refractive errors

Methods: A double-blind study was conducted to evaluate accuracy of a prototype low-cost wavefront aberrometer in obtaining a refractive prescription. The prototype was compared with a Grand Seiko WR-5100K open-field autorefractor and binocular subjective refraction. The prototype is a handheld, lightweight, open-view, and easy-to-use device that quickly provides a refractive prescription without requiring pupil dilation. The prototype includes a Hartmann-Shack lenslet array with a low-cost CMOS image sensor for wavefront sensing, and an 850 nm laser diode for illumination (maximum corneal power of 250 μW). All the functional components of the device are off-the-shelf parts that cost less than $1,000 in total. The subjects held and looked through the prototype while a 30-second video of spot diagrams was captured. The spot diagrams were processed using a custom algorithm to calculate Zernike coefficients and estimate a prescription. Refractions were obtained for 43 subjects (mean age 26.2 ± 9.5 years) with each method. Eight of these subjects were used for validation and development of the prototype algorithm, and the remaining 35 were used for the double-blind test.

Results: For the 35 subjects, the range of spherical equivalent (SE) refractive error measured by subjective refraction was -6.50 to +3.63 D. The correlations between the SE measured objectively and subjectively were R = 0.96 for the prototype and R = 0.97 for the autorefractor. The average SE error of the prototype compared to subjective refraction was 0.54 ± 0.54 D, versus 0.40 ± 0.46 D for the autorefractor. The average errors of the J0 and J45 power vectors were 0.16 ± 0.22 D and 0.11 ± 0.11 D for the prototype and 0.13 ± 0.09 D and 0.08 ± 0.07 D for the autorefractor compared to subjective refraction.

Conclusions: The prototype wavefront aberrometer performed similarly to a high-end open-field commercial autorefractor in objective refraction when using subjective refraction as the gold standard. The prototype performed worst on subjects with anisometropia (n=2), likely due to cross-coupled accommodation. Future improvements in the prototype algorithm can improve its accuracy in measuring refractive errors. An easy-to-use and low-cost autorefractor, such as the one evaluated here, may be beneficial for improving eye care in low-resource settings.