Refractive Surgery/Vision Correction

Deep Learning Algorithms May be Beneficial in Screening Candidates for Refractive Surgery

Posted on

In this diagnostic, cross-sectional study, 6465 corneal tomographic images from 1385 patients were examined. Pentacam InceptionResNetV2 Screening System (PIRSS), a tomographic-based screening tool, was developed to screen potential patients for refractive surgery. On the validation data set, the model achieved an overall detection accuracy of 94.7%. On the independent test data set, the model achieved an overall detection accuracy of 95%, which is on par with a senior ophthalmologist who specializes in refractive surgery (92.8%).

The model also performed better than classifiers in identifying corneas with contraindications for refractive surgery (95% vs 81%; P  < 0.001) in this population.

The authors concluded that the PRISS model appears to be useful in providing corneal information and identifying at-risk patients and may provide guidance to refractive surgeon in screening candidates.

Read the full article here.



Xie Y, Zhao L, Yang X, et al. Screening candidates for refractive surgery with corneal tomographic–based deep learning. JAMA Ophthalmol. Published online March 26, 2020. doi:10.1001/jamaophthalmol.2020.0507

Related Articles
Which is a better preventive pain regimen after photorefractive keratectomy: oral or topical NSAIDs?
Jun 03, 2020
Determinants of Patient Satisfaction in Post-LASIK Patients Undergoing Cataract Surgery
May 27, 2020
How is the CenterVue Compass Visual Field Different?
Jan 21, 2020