Search

PurposeTo develop and evaluate an automated, portable algorithm to differentiate active corneal ulcers from healed scars using only external photographs.DesignA convolutional neural network was trained and tested using photographs of corneal ulcers and scars.ParticipantsDe-identified photographs of corneal ulcers were obtained from the Steroids for Corneal Ulcers Trial (SCUT), Mycotic Ulcer Treatment Trial (MUTT), and Byers Eye Institute at Stanford University.MethodsPhotographs of corneal ulcers (n = 1313) and scars (n = 1132) from the SCUT and MUTT were used to train a convolutional neural network (CNN). The CNN was tested on 2 different patient populations from eye clinics in India (n = 200) and the Byers Eye Institute at Stanford University (n = 101). Accuracy was evaluated against gold standard clinical classifications. Feature importances for the trained model were visualized using gradient-weighted class activation mapping.Main outcome measuresAccuracy of the CNN was assessed via F1 score. The area under the receiver operating characteristic (ROC) curve (AUC) was used to measure the precision-recall trade-off.ResultsThe CNN correctly classified 115 of 123 active ulcers and 65 of 77 scars in patients with corneal ulcer from India (F1 score, 92.0% [95% confidence interval (CI), 88.2%-95.8%]; sensitivity, 93.5% [95% CI, 89.1%-97.9%]; specificity, 84.42% [95% CI, 79.42%-89.42%]; ROC: AUC, 0.9731). The CNN correctly classified 43 of 55 active ulcers and 42 of 46 scars in patients with corneal ulcers from Northern California (F1 score, 84.3% [95% CI, 77.2%-91.4%]; sensitivity, 78.2% [95% CI, 67.3%-89.1%]; specificity, 91.3% [95% CI, 85.8%-96.8%]; ROC: AUC, 0.9474). The CNN visualizations correlated with clinically relevant features such as corneal infiltrate, hypopyon, and conjunctival injection.ConclusionsThe CNN classified corneal ulcers and scars with high accuracy and generalized to patient populations outside of its training data. The CNN f

Natacha C Villegas, Wen-Shin Lee Department of Ophthalmology, Stanford University School of Medicine, Stanford, CA, USACorrespondence: Natacha C VillegasDepartment of Ophthalmology, Stanford University, Medical Center, 2452 Watson Court, Palo Alto, CA, 94303-5353, USATel +1 832 477-3498Email natachav@stanford.eduPurpose: This study seeks to evaluate the effectiveness of netarsudil (Rhopressa) in patients with inadequately controlled IOP on otherwise maximally tolerated medical therapy.Methods: This is a retrospective study of patients started on netarsudil at Stanford University. Exclusion criteria included glaucoma surgery or laser within 6 months of starting netarsudil and other modifications to the baseline medication regimen within 4 weeks of starting netarsudil. The primary outcome was treatment success, defined as IOP reduction meeting a predetermined target, and no further medication, laser, or surgery recommended subsequent to starting netarsudil.Results: Sixty-two eyes were included, and 36 (58%) achieved treatment success at first follow-up. Mean baseline IOP was 19.5 ± 5.6 mmHg on a mean of 3.5 ± 0.7 ocular hypotensive medications. The mean change in IOP from baseline to first follow-up was − 3.53 mmHg (− 17%). In patients who achieved treatment success, mean IOP change was − 5.22 mmHg (− 28.0%). Of the eyes with baseline IOP ≤ 20 mmHg, 69% achieved treatment success, compared to only 17% of eyes with baseline IOP ≥ 21 mmHg (P < 0.05).Conclusion: Netarsudil is effective in lowering IOP for patients on otherwise maximally tolerated medical therapy, for which glaucoma laser or surgery would have been the only remaining therapeutic options. Treatment success was more likely in eyes with baseline IOP under 20 mmHg.Keywords: netarsudil, Rhopressa, maximally tolerated medical therapy, glaucoma