1. Optimized Artificial Intelligence for Enhanced Ectasia Detection Using Scheimpflug-Based Corneal Tomography and Biomechanical Data
- Author
-
Renato Ambrósio, Aydano P. Machado, Edileuza Leão, João Marcelo G. Lyra, Marcella Q. Salomão, Louise G. Pellegrino Esporcatte, João B.R. da Fonseca Filho, Erica Ferreira-Meneses, Nelson B. Sena, Jorge S. Haddad, Alexandre Costa Neto, Gildasio Castelo de Almeida, Cynthia J. Roberts, Ahmed Elsheikh, Riccardo Vinciguerra, Paolo Vinciguerra, Jens Bühren, Thomas Kohnen, Guy M. Kezirian, Farhad Hafezi, Nikki L. Hafezi, Emilio A. Torres-Netto, Nanji Lu, David Sung Yong Kang, Omid Kermani, Shizuka Koh, Prema Padmanabhan, Suphi Taneri, William Trattler, Luca Gualdi, José Salgado-Borges, Fernando Faria-Correia, Elias Flockerzi, Berthold Seitz, Vishal Jhanji, Tommy C.Y. Chan, Pedro Manuel Baptista, Dan Z. Reinstein, Timothy J. Archer, Karolinne M. Rocha, George O. Waring, Ronald R. Krueger, William J. Dupps, Ramin Khoramnia, Hassan Hashemi, Soheila Asgari, Hamed Momeni-Moghaddam, Siamak Zarei-Ghanavati, Rohit Shetty, Pooja Khamar, Michael W. Belin, and Bernardo T. Lopes
- Subjects
Ophthalmology ,Human medicine - Abstract
center dot PURPOSE: To optimize artificial intelligence (AI) algo-rithms to integrate Scheimpflug-based corneal tomogra-phy and biomechanics to enhance ectasia detection.center dot DESIGN: Multicenter cross-sectional case-control ret-rospective study.center dot METHODS: A total of 3886 unoperated eyes from 3412 patients had Pentacam and Corvis ST (Oculus Op-tikgerate GmbH) examinations. The database included 1 eye randomly selected from 1680 normal patients (N) and from 1181 "bilateral" keratoconus (KC) patients, along with 551 normal topography eyes from patients with very asymmetric ectasia (VAE-NT), and their 474 unoperated ectatic (VAE-E) eyes. The current TBIv1 (tomographic-biomechanical index) was tested, and an optimized AI algorithm was developed for augmenting accuracy.center dot RESULTS: The area under the receiver operating char-acteristic curve (AUC) of the TBIv1 for discriminating clinical ectasia (KC and VAE-E) was 0.999 (98.5% sen-sitivity; 98.6% specificity [cutoff: 0.5]), and for VAE-NT, 0.899 (76% sensitivity; 89.1% specificity [cutoff: 0.29]). A novel random forest algorithm (TBIv2), devel-oped with 18 features in 156 trees using 10-fold cross -validation, had a significantly higher AUC (0.945; De -Long, P < .0001) for detecting VAE-NT (84.4% sen-sitivity and 90.1% specificity; cutoff: 0.43; DeLong, P < .0001) and a similar AUC for clinical ectasia (0.999; DeLong, P = .818; 98.7% sensitivity; 99.2% specificity [cutoff: 0.8]). Considering all cases, the TBIv2 had a higher AUC (0.985) than TBIv1 (0.974; DeLong, P < .0001).center dot CONCLUSIONS: AI optimization to integrate Scheimpflug-based corneal tomography and biome-chanical assessments augments accuracy for ectasia detection, characterizing ectasia susceptibility in the diverse VAE-NT group. Some patients with VAE may have true unilateral ectasia. Machine learning consider-ing additional data, including epithelial thickness or other parameters from multimodal refractive imaging, will con-tinuously enhance accuracy. NOTE: Publication of this article is sponsored by the American Ophthalmological Society. (Am J Ophthalmol 2023;251: 126-142. (c) 2023 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ))
- Published
- 2023
- Full Text
- View/download PDF