Calculating intraocular lens geometry by real ray tracing

An implementation of real ray tracing based on Snell's law is tested by predicting the refraction of pseudophakic eyes and calculating the geometry of intraocular lenses (IOLs).

The refraction of 30 pseudophakic eyes was predicted with the measured corneal topography, axial length, and the known IOL geometry and compared to the manifest refraction. Intraocular lens calculation was performed for 30 normal eyes and 12 eyes that had previous refractive surgery for myopia correction and compared to state-of-the-art IOL calculation formulae.

Mean difference between predicted and manifest refraction for a 2.5-mm pupil were sphere 0.11 +/- 0.43 diopters (D), cylinder -0.18 +/- 0.52 D, and axis 5.13 degrees +/- 30.19 degrees. Pearson's correlation coefficient was sphere r = 0.92, P < .01; cylinder r = 0.79, P < .01; and axis r = 0.91, P < .01. Intraocular lens calculation for the normal group showed that the mean absolute error regarding refractive outcome is largest for SRK II (0.49 D); all other formulae including ray tracing result in similar values ranging from 0.36 to 0.40 D. Intraocular lens calculation for the refractive group showed that depending on pupil size (3.5 to 2.5 mm), ray tracing delivers values 0.95 to 1.90 D higher compared to the average of Holladay 1, SRK/T, Haigis, and Hoffer Q formulae.

It has been shown that ray tracing can compete with state-of-the-art IOL calculation formulae for normal eyes. For eyes with previous refractive surgery, IOL powers obtained by ray tracing are significantly higher than those from the other formulae. Thus, a hyperopic shift may be avoided using ray tracing even without clinical history.