@article{CasadoMorenoBOEx2025,
author = {Juan Casado-Moreno and Belen Masia and Nanji Lu and Lele Cui and Alejandra Consejo},
journal = {Biomed. Opt. Express},
keywords = {Deep learning; Imaging techniques; Laser surgery; Machine learning; Spatial resolution; Three dimensional imaging},
number = {8},
pages = {3047--3060},
publisher = {Optica Publishing Group},
title = {Deep learning-based keratoconus detection from Scheimpflug images},
volume = {16},
month = {Aug},
year = {2025},
url = {https://opg.optica.org/boe/abstract.cfm?URI=boe-16-8-3047},
doi = {10.1364/BOE.559663},
abstract = {This study evaluates the effectiveness of deep learning techniques applied to raw Scheimpflug corneal images for keratoconus detection, with a particular focus on forme fruste (FF) keratoconus, which refers to preclinical cases. Using an original dataset of 22,750 images from 910 eyes, a deep learning model based on transfer learning with a pre-trained VGG16 architecture was trained, incorporating specific preprocessing steps and data augmentation strategies. The proposed approach achieved an overall accuracy of 90.70\&\#x0025;, with a sensitivity of 80.57\&\#x0025;, and a specificity of 80.56\&\#x0025; for FF keratoconus classification, and an AUC of 0.89. For clinical keratoconus, the model demonstrated a sensitivity of 93.28\&\#x0025;, a specificity of 99.40\&\#x0025;, and an AUC of 1.00. These findings highlight the potential of leveraging raw Scheimpflug images in deep learning-based keratoconus detection, particularly for identifying early-stage structural changes that may not be apparent in conventional topographic assessments.},
}