AI-powered effective lens position prediction improves the accuracy of existing lens formulas

AimsTo assess whether incorporating a machine learning (ML) method for accurate prediction of postoperative anterior chamber depth (ACD) improves the refraction prediction performance of existing intraocular lens (IOL) calculation formulas.MethodsA dataset of 4806 patients with cataract was gathered at the Kellogg Eye Center, University of Michigan, and split into a training set (80% of patients, 5761 eyes) and a testing set (20% of patients, 961 eyes). A previously developed ML-based method was used to predict the postoperative ACD based on preoperative biometry. This ML-based postoperative ACD was integrated into new effective lens position (ELP) predictions using regression models to rescale the ML output for each of four existing formulas (Haigis, Hoffer Q, Holladay and SRK/T). The performance of the formulas with ML-modified ELP was compared using a testing dataset. Performance was measured by the mean absolute error (MAE) in refraction prediction.ResultsWhen the ELP was replaced with a linear combination of the original ELP and the ML-predicted ELP, the MAEs±SD (in Diopters) in the testing set were: 0.356±0.329 for Haigis, 0.352±0.319 for Hoffer Q, 0.371±0.336 for Holladay, and 0.361±0.331 for SRK/T which were significantly lower (p<0.05) than those of the original formulas: 0.373±0.328 for Haigis, 0.408±0.337 for Hoffer Q, 0.384±0.341 for Holladay and 0.394±0.351 for SRK/T.ConclusionUsing a more accurately predicted postoperative ACD significantly improves the prediction accuracy of four existing IOL power formulas.

Refractive Error Induced by Intraocular Lens Tilt After Intrascleral Intraocular Lens Fixation

Purpose To investigate the spherical shift of Intraocular lens (IOL) tilt after intrascleral fixationMethods We retrospectively reviewed the medical records of patients who underwent flanged intrascleral IOL fixation with transconjunctival 25- or 27-gauge pars plana vitrectomy at the Department of Ophthalmology of Jikei University Hospital with a minimum follow-up duration of three months. Second-generation anterior segment optical coherence tomography (CASIA2; TOMEY) was used to obtain the tilt and decentration of intrasclerally fixated IOL and postoperative anterior chamber depth. We investigated the relationship of the refractive error with these parameters, axial length, and keratometry. In addition to the clinical investigation, we performed optical simulations using the Zemax optical design program for studying the spherical shift of the IOL tilt by means of the through-focus response and change of spherical equivalent power.Results The study involved 72 eyes of 67 patients. The degree of IOL tilt was correlated with the amount of refractive error (Spearman's rank correlation coefficient [CC] = −0.32; P = 0.006). In particular, a tilt angle greater than 10° strongly influenced the refractive error. Postoperative anterior chamber depth also correlated with the refractive error (CC = 0.50; P < 0.001). The refractive error did not correlate with decentration (CC = −0.17; P = 0.15), axial length (CC = −0.08; P = 0.49), and keratometry (CC = −0.06; P =0.64). Optical simulations using the Zemax optical design program also showed a myopic shift exponentially as the tilt becomes greater. Conclusion An IOL tilt greater than 10 ° induces refractive error.

AI-Powered Effective Lens Position Prediction Improves the Accuracy of Existing Lens Formulas

ABSTRACTAimsTo assess whether incorporating a machine learning (ML) method for accurate prediction of postoperative anterior chamber depth (ACD) improves the refraction prediction performance of existing intraocular lens (IOL) calculation formulas.MethodsA dataset of 4806 cataract patients were gathered at the Kellogg Eye Center, University of Michigan, and split into a training set (80% of patients, 5761 eyes) and a testing set (20% of patients, 961 eyes). A previously developed ML-based method was used to predict the postoperative ACD based on preoperative biometry. This ML-based postoperative ACD was integrated into new effective lens position (ELP) predictions using regression models to rescale the ML output for each of four existing formulas (Haigis, Hoffer Q, Holladay, and SRK/T). The performance of the formulas with ML-modified ELP was compared using a testing dataset. Performance was measured by the mean absolute error (MAE) in refraction prediction.ResultsWhen the ELP was replaced with a linear combination of the original ELP and the ML-predicted ELP, the MAEs ± SD (in Diopters) in the testing set were: 0.356 ± 0.329 for Haigis, 0.352 ± 0.319 for Hoffer Q, 0.371 ± 0.336 for Holladay, and 0.361 ± 0.331 for SRK/T which were significantly lower than those of the original formulas: 0.373 ± 0.328 for Haigis, 0.408 ± 0.337 for Hoffer Q, 0.384 ± 0.341 for Holladay, and 0.394 ± 0.351 for SRK/T.ConclusionUsing a more accurately predicted postoperative ACD significantly improves the prediction accuracy of four existing IOL power formulas.