Differentiation of Active Corneal Infections From Healed Scars Using Deep Learning.

Abstract

OBJECTIVE\nTo develop and evaluate an automated, portable algorithm to differentiate active corneal ulcers from healed scars using only external photographs.\n\n\nDESIGN\nA convolutional neural network was trained and tested using photographs of corneal ulcers and scars.\n\n\nSUBJECTS\nDe-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.\n\n\nMETHODS\nPhotographs of corneal ulcers (n=1313) and scars (n=1132) from the SCUT and MUTT trials were used to train a convolutional neural network (CNN). The CNN was tested on two 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 (Grad-CAM).\n\n\nMAIN OUTCOME MEASURE\nAccuracy of the CNN was assessed via F1 score. Area under the receiver operating characteristic curve (ROC) was used to measure the precision-recall trade-off.\n\n\nRESULTS\nThe CNN correctly classified 115/123 active ulcers and 65/77 scars in corneal ulcer patients from India (F1 score: 92.0% (95% 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 43/55 active ulcers and 42/46 scars in corneal ulcer patients 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.\n\n\nCONCLUSION\nThe CNN classified corneal ulcers and scars with high accuracy and generalizes to patient populations outside of its training data. The CNN focuses on clinically relevant features when it makes a diagnosis. The CNN demonstrates potential as an inexpensive diagnostic approach that may aid triage in communities with limited access to eye care.