Mo Tiwari1, Chris Piech1, Medina Baitemirova2, Namperumalsamy V Prajna3, Muthiah Srinivasan3, Prajna Lalitha3, Natacha Villegas4, Niranjan Balachandar1, Janice T Chua5, Travis Redd6, Thomas M Lietman7, Sebastian Thrun1, Charles C Lin8. 1. Department of Computer Science, Stanford University, Stanford, California. 2. Department of Biomedical Informatics, Stanford University, Stanford, California. 3. Aravind Eye Hospital, Madurai, India. 4. Byers Eye Institute, Stanford University, Stanford, California. 5. School of Medicine, University of California, Irvine, Irvine, California. 6. Department of Ophthalmology, Casey Eye Institute, Oregon Health and Science University, Portland, Oregon. 7. Francis I. Proctor Foundation, University of California San Francisco, San Francisco, California. 8. Byers Eye Institute, Stanford University, Stanford, California. Electronic address: lincc@stanford.edu.
Abstract
PURPOSE: To develop and evaluate an automated, portable algorithm to differentiate active corneal ulcers from healed scars using only external photographs. DESIGN: A convolutional neural network was trained and tested using photographs of corneal ulcers and scars. PARTICIPANTS: De-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. METHODS: Photographs 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 MEASURES: Accuracy 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. RESULTS: The 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. CONCLUSIONS: The CNN classified corneal ulcers and scars with high accuracy and generalized to patient populations outside of its training data. The CNN focused on clinically relevant features when it made a diagnosis. The CNN demonstrated potential as an inexpensive diagnostic approach that may aid triage in communities with limited access to eye care.
PURPOSE: To develop and evaluate an automated, portable algorithm to differentiate active corneal ulcers from healed scars using only external photographs. DESIGN: A convolutional neural network was trained and tested using photographs of corneal ulcers and scars. PARTICIPANTS: De-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. METHODS: Photographs 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 MEASURES: Accuracy 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. RESULTS: The 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. CONCLUSIONS: The CNN classified corneal ulcers and scars with high accuracy and generalized to patient populations outside of its training data. The CNN focused on clinically relevant features when it made a diagnosis. The CNN demonstrated potential as an inexpensive diagnostic approach that may aid triage in communities with limited access to eye care.
Authors: Maria A Woodward; David C Musch; Christopher T Hood; Jonathan B Greene; Leslie M Niziol; V Swetha E Jeganathan; Paul P Lee Journal: Cornea Date: 2017-10 Impact factor: 2.651
Authors: John Zech; Margaret Pain; Joseph Titano; Marcus Badgeley; Javin Schefflein; Andres Su; Anthony Costa; Joshua Bederson; Joseph Lehar; Eric Karl Oermann Journal: Radiology Date: 2018-01-30 Impact factor: 11.105
Authors: Andre Esteva; Brett Kuprel; Roberto A Novoa; Justin Ko; Susan M Swetter; Helen M Blau; Sebastian Thrun Journal: Nature Date: 2017-01-25 Impact factor: 49.962
Authors: Bennie H Jeng; David C Gritz; Abha B Kumar; Douglas S Holsclaw; Travis C Porco; Scott D Smith; John P Whitcher; Todd P Margolis; Ira G Wong Journal: Arch Ophthalmol Date: 2010-08
Authors: N Venkatesh Prajna; Tiruvengada Krishnan; Jeena Mascarenhas; Revathi Rajaraman; Lalitha Prajna; Muthiah Srinivasan; Anita Raghavan; Catherine E Oldenburg; Kathryn J Ray; Michael E Zegans; Stephen D McLeod; Travis C Porco; Nisha R Acharya; Thomas M Lietman Journal: JAMA Ophthalmol Date: 2013-04 Impact factor: 7.389
Authors: Charles C Lin; Prajna Lalitha; Muthiah Srinivasan; N Venkatesh Prajna; Stephen D McLeod; Nisha R Acharya; Thomas M Lietman; Travis C Porco Journal: Cornea Date: 2012-10 Impact factor: 2.651
Authors: Philippe M Burlina; Neil Joshi; Michael Pekala; Katia D Pacheco; David E Freund; Neil M Bressler Journal: JAMA Ophthalmol Date: 2017-11-01 Impact factor: 7.389
Authors: Darren Shu Jeng Ting; Mohamed Galal; Bina Kulkarni; Mohamed S Elalfy; Damian Lake; Samer Hamada; Dalia G Said; Harminder S Dua Journal: J Fungi (Basel) Date: 2021-11-12