Deep Learning Model for Static Ocular Torsion Detection Using Synthetically Generated Fundus Images
Translational Vision Science & Technology · 11.01.2023 · 12 (1) · 17
DOI: 10.1167/tvst.12.1.17 ↗Kurz gefasst
Fachartikel in der Zeitschrift Translational Vision Science & Technology, erschienen am 11.01.2023. Verfasst von Chen Wang und 15 weiteren Autor:innen. Beteiligte Einrichtung: Johns Hopkins University Department of Biomedical Engineering, Baltimore, MD, USA. Hauptthemen laut MEDLINE: Deep Learning. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 55 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
Purpose
The objective of the study is to develop deep learning models using synthetic fundus images to assess the direction (intorsion versus extorsion) and amount (physiologic versus pathologic) of static ocular torsion. Static ocular torsion assessment is an important clinical tool for classifying vertical ocular misalignment; however, current methods are time-intensive with steep learning curves for frontline providers.
Methods
We used a dataset (n = 276) of right eye fundus images. The disc-foveal angle was calculated using ImageJ to generate synthetic images via image rotation. Using synthetic datasets (n = 12,740 images per model) and transfer learning (the reuse of a pretrained deep learning model on a new task), we developed a binary classifier (intorsion versus extorsion) and a multiclass classifier (physiologic versus pathologic intorsion and extorsion). Model performance was evaluated on unseen synthetic and nonsynthetic data.
Results
On the synthetic dataset, the binary classifier had an accuracy and area under the receiver operating characteristic curve (AUROC) of 0.92 and 0.98, respectively, whereas the multiclass classifier had an accuracy and AUROC of 0.77 and 0.94, respectively. The binary classifier generalized well on the nonsynthetic data (accuracy = 0.94; AUROC = 1.00).
Conclusions
The direction of static ocular torsion can be detected from synthetic fundus images using deep learning methods, which is key to differentiate between vestibular misalignment (skew deviation) and ocular muscle misalignment (superior oblique palsies). Translational Relevance: Given the robust performance of our models on real fundus images, similar strategies can be adopted for deep learning research in rare neuro-ophthalmologic diseases with limited datasets.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Johns Hopkins University Department of Biomedical Engineering, Baltimore, MD, USA.
- Johns Hopkins University School of Medicine, Department of Neurology, Baltimore, MD, USA.
- University of California Berkeley, Herbert Wertheim School of Optometry and Vision Science, Berkeley, CA, USA.
- Johns Hopkins University School of Medicine, Department of Ophthalmology, Baltimore, MD, USA.
Zitieren
APA
Wang, C., Bai, Y., Tsang, A., Bian, Y., Gou, Y., Lin, Y. X., Zhao, M., Wei, T. Y., Desman, J. M., Taylor, C. O., Greenstein, J. L., Otero-Millan, J., Liu, T. Y. A., Kheradmand, A., Zee, D. S., & Green, K. E. (2023). Deep Learning Model for Static Ocular Torsion Detection Using Synthetically Generated Fundus Images. Translational Vision Science & Technology, 12(1), 17. https://doi.org/10.1167/tvst.12.1.17
Vancouver
Wang C, Bai Y, Tsang A, Bian Y, Gou Y, Lin YX, et al. Deep Learning Model for Static Ocular Torsion Detection Using Synthetically Generated Fundus Images. Translational Vision Science & Technology. 2023;12(1):17. doi:10.1167/tvst.12.1.17
BibTeX
@article{Wang2023,
author = {Wang, Chen and Bai, Yunong and Tsang, Ashley and Bian, Yuhan and Gou, Yifan and Lin, Yan X and Zhao, Matthew and Wei, Tony Y and Desman, Jacob M and Taylor, Casey Overby and Greenstein, Joseph L and Otero-Millan, Jorge and Liu, Tin Yan Alvin and Kheradmand, Amir and Zee, David S and Green, Kemar E},
title = {Deep Learning Model for Static Ocular Torsion Detection Using Synthetically Generated Fundus Images},
journal = {Translational Vision Science & Technology},
year = {2023},
volume = {12},
number = {1},
pages = {17},
doi = {10.1167/tvst.12.1.17}
}
EUVIMED