Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study

Purpose To develop an artificial intelligence (AI) approach for differentiating between normal cornea, subclinical, and keratoconus (KC) using tomographic maps from Pentacam (Oculus) and corneal biomechanics from Corvis ST (Oculus). Methods A total of 1,668 tomographic (769 patients) and 611 biomech...

Full description

Saved in:
Bibliographic Details
Main Authors: Wiyada Quanchareonsap, Ngamjit Kasetsuwan, Usanee Reinprayoon, Yonrawee Piyacomn, Thitima Wungcharoen, Monthira Jermjutitham
Format: Article
Language:English
Published: Wolters Kluwer Medknow Publications 2024-10-01
Series:Journal of Current Ophthalmology
Subjects:
Online Access:https://journals.lww.com/10.4103/joco.joco_18_24
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850199609549258752
author Wiyada Quanchareonsap
Ngamjit Kasetsuwan
Usanee Reinprayoon
Yonrawee Piyacomn
Thitima Wungcharoen
Monthira Jermjutitham
author_facet Wiyada Quanchareonsap
Ngamjit Kasetsuwan
Usanee Reinprayoon
Yonrawee Piyacomn
Thitima Wungcharoen
Monthira Jermjutitham
author_sort Wiyada Quanchareonsap
collection DOAJ
description Purpose To develop an artificial intelligence (AI) approach for differentiating between normal cornea, subclinical, and keratoconus (KC) using tomographic maps from Pentacam (Oculus) and corneal biomechanics from Corvis ST (Oculus). Methods A total of 1,668 tomographic (769 patients) and 611 biomechanical (307 patients) images from the Chula Refractive Surgery Center, King Chulalongkorn Memorial Hospital were included. The sample size was divided into the Pentacam and combined Pentacam-Corvis groups. Different convolutional neural network approaches were used to enhance the KC and subclinical KC detection performance. Results AI model 1, which obtained refractive maps from Pentacam, achieved an area under the receiver operating characteristic curve (AUC) of 0.938 and accuracy of 0.947 (sensitivity, 90.8% and specificity, 96.9%). AI model 2, which added dynamic corneal response and the Vinciguerra screening report from Corvis ST to AI Model 1, achieved an AUC of 0.985 and accuracy of 0.956 (sensitivity, 93.0% and specificity, 94.3%). AI model 3, which added the corneal biomechanical index to AI Model 2, reached an AUC of 0.991 and accuracy of 0.956 (sensitivity, 93.0% and specificity, 94.3%). Conclusions Our study showed that AI models using either anterior corneal curvature alone or combined with corneal biomechanics could help classify normal and keratoconic corneas, which would make diagnosis more accurate and would be helpful in decision-making for the treatment.
format Article
id doaj-art-e748f6f10fef4c5f93cf7cb456de3062
institution OA Journals
issn 2452-2325
language English
publishDate 2024-10-01
publisher Wolters Kluwer Medknow Publications
record_format Article
series Journal of Current Ophthalmology
spelling doaj-art-e748f6f10fef4c5f93cf7cb456de30622025-08-20T02:12:34ZengWolters Kluwer Medknow PublicationsJournal of Current Ophthalmology2452-23252024-10-01361465310.4103/joco.joco_18_24Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic StudyWiyada QuanchareonsapNgamjit KasetsuwanUsanee ReinprayoonYonrawee PiyacomnThitima WungcharoenMonthira JermjutithamPurpose To develop an artificial intelligence (AI) approach for differentiating between normal cornea, subclinical, and keratoconus (KC) using tomographic maps from Pentacam (Oculus) and corneal biomechanics from Corvis ST (Oculus). Methods A total of 1,668 tomographic (769 patients) and 611 biomechanical (307 patients) images from the Chula Refractive Surgery Center, King Chulalongkorn Memorial Hospital were included. The sample size was divided into the Pentacam and combined Pentacam-Corvis groups. Different convolutional neural network approaches were used to enhance the KC and subclinical KC detection performance. Results AI model 1, which obtained refractive maps from Pentacam, achieved an area under the receiver operating characteristic curve (AUC) of 0.938 and accuracy of 0.947 (sensitivity, 90.8% and specificity, 96.9%). AI model 2, which added dynamic corneal response and the Vinciguerra screening report from Corvis ST to AI Model 1, achieved an AUC of 0.985 and accuracy of 0.956 (sensitivity, 93.0% and specificity, 94.3%). AI model 3, which added the corneal biomechanical index to AI Model 2, reached an AUC of 0.991 and accuracy of 0.956 (sensitivity, 93.0% and specificity, 94.3%). Conclusions Our study showed that AI models using either anterior corneal curvature alone or combined with corneal biomechanics could help classify normal and keratoconic corneas, which would make diagnosis more accurate and would be helpful in decision-making for the treatment.https://journals.lww.com/10.4103/joco.joco_18_24artificial intelligenceforme fruste keratoconuskeratoconuskeratoconus suspectmachine learningsubclinical keratoconus
spellingShingle Wiyada Quanchareonsap
Ngamjit Kasetsuwan
Usanee Reinprayoon
Yonrawee Piyacomn
Thitima Wungcharoen
Monthira Jermjutitham
Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
Journal of Current Ophthalmology
artificial intelligence
forme fruste keratoconus
keratoconus
keratoconus suspect
machine learning
subclinical keratoconus
title Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
title_full Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
title_fullStr Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
title_full_unstemmed Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
title_short Deep Learning Algorithm for Keratoconus Detection from Tomographic Maps and Corneal Biomechanics: A Diagnostic Study
title_sort deep learning algorithm for keratoconus detection from tomographic maps and corneal biomechanics a diagnostic study
topic artificial intelligence
forme fruste keratoconus
keratoconus
keratoconus suspect
machine learning
subclinical keratoconus
url https://journals.lww.com/10.4103/joco.joco_18_24
work_keys_str_mv AT wiyadaquanchareonsap deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy
AT ngamjitkasetsuwan deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy
AT usaneereinprayoon deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy
AT yonraweepiyacomn deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy
AT thitimawungcharoen deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy
AT monthirajermjutitham deeplearningalgorithmforkeratoconusdetectionfromtomographicmapsandcornealbiomechanicsadiagnosticstudy