1. Detection of glaucoma progression on longitudinal series of en-face macular optical coherence tomography angiography images with a deep learning model.
- Author
-
Mohammadzadeh V, Liang Y, Moghimi S, Xie P, Nishida T, Mahmoudinezhad G, Eslani M, Walker E, Kamalipour A, Micheletti E, Wu JH, Christopher M, Zangwill LM, Javidi T, and Weinreb RN
- Subjects
- Humans, Female, Male, Middle Aged, Aged, Fluorescein Angiography methods, ROC Curve, Follow-Up Studies, Optic Disk diagnostic imaging, Optic Disk blood supply, Optic Disk pathology, Macula Lutea diagnostic imaging, Macula Lutea pathology, Retinal Ganglion Cells pathology, Visual Field Tests, Retinal Vessels diagnostic imaging, Retinal Vessels pathology, Retrospective Studies, Tomography, Optical Coherence methods, Deep Learning, Disease Progression, Visual Fields physiology, Glaucoma, Open-Angle physiopathology, Glaucoma, Open-Angle diagnosis, Glaucoma, Open-Angle diagnostic imaging, Intraocular Pressure physiology
- Abstract
Background/aims: To design a deep learning (DL) model for the detection of glaucoma progression with a longitudinal series of macular optical coherence tomography angiography (OCTA) images., Methods: 202 eyes of 134 patients with open-angle glaucoma with ≥4 OCTA visits were followed for an average of 3.5 years. Glaucoma progression was defined as having a statistically significant negative 24-2 visual field (VF) mean deviation (MD) rate. The baseline and final macular OCTA images were aligned according to centre of fovea avascular zone automatically, by checking the highest value of correlation between the two images. A customised convolutional neural network (CNN) was designed for classification. A comparison of the CNN to logistic regression model for whole image vessel density (wiVD) loss on detection of glaucoma progression was performed. The performance of the model was defined based on the confusion matrix of the validation dataset and the area under receiver operating characteristics (AUC)., Results: The average (95% CI) baseline VF MD was -3.4 (-4.1 to -2.7) dB. 28 (14%) eyes demonstrated glaucoma progression. The AUC (95% CI) of the DL model for the detection of glaucoma progression was 0.81 (0.59 to 0.93). The sensitivity, specificity and accuracy (95% CI) of DL model were 67% (34% to 78%), 83% (42% to 97%) and 80% (52% to 95%), respectively. The AUC (95% CI) for the detection of glaucoma progression based on the logistic regression model was lower than the DL model (0.69 (0.50 to 0.88))., Conclusion: The optimised DL model detected glaucoma progression based on longitudinal macular OCTA images showed good performance. With external validation, it could enhance detection of glaucoma progression., Trial Registration Number: NCT00221897., Competing Interests: Competing interests: VM: None; YL: None; SM: F: National Eye Institute; PX: None; TN: C: Topcon; GM: None; ME: None; EW: None; AK: F: Fight for Sight; EM: None; J-HW: None:, MC: F: National Eye Institute; LMZ: C: Abbvie Inc., Topcon; F: National Eye Institute, Carl Zeiss Meditec Inc., Heidelberg Engineering GmbH, Optovue Inc., Topcon Medical Systems Inc.; P: Zeiss Meditec, AISight Health (founder); TJ: None; RNW: C: Abbvie, Aerie Pharmaceuticals, Allergan, Amydis, Editas, Equinox, Eyenovia, Iantrek, Implandata, IOPtic, iSTAR Medical, Nicox, Santen, Tenpoint and Topcon; F: National Eye Institute, National Institute of Minority Health and Health Disparities, Heidelberg Engineering, Carl Zeiss Meditec, Konan Medical, Optovue, Zilia, Centervue, and Topcon; P: Toromedes, Carl Zeiss Meditec., (© Author(s) (or their employer(s)) 2024. No commercial re-use. See rights and permissions. Published by BMJ.)
- Published
- 2024
- Full Text
- View/download PDF