June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Combining OCT and OCT-Angiography Longitudinal Measurements for the Prediction of Visual Field Progression in Glaucoma with Artificial Intelligence
Author Affiliations & Notes
  • Alireza Kamalipour
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Sasan Moghimi
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Pooya Khosravi
    University of California Irvine, Irvine, California, United States
  • Vahid Mohammadzadeh
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Takashi Nishida
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Eleonora Micheletti
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Jo-Hsuan Wu
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Golnoush Mahmoudinezhad
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Mark Christopher
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Linda Zangwill
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Tara Javidi
    Electrical and Computer Engineering, University of California San Diego, La Jolla, California, United States
  • Robert N Weinreb
    Ophthalmology, University of California San Diego, La Jolla, California, United States
  • Footnotes
    Commercial Relationships   Alireza Kamalipour None; Sasan Moghimi None; Pooya Khosravi None; Vahid Mohammadzadeh None; Takashi Nishida None; Eleonora Micheletti None; Jo-Hsuan Wu None; Golnoush Mahmoudinezhad None; Mark Christopher National Eye Institute, Code F (Financial Support); Linda Zangwill Abbvie and Digital Diagnostics, Code C (Consultant/Contractor), National Eye Institute, Code F (Financial Support), Carl Zeiss Meditec Inc., Code F (Financial Support), Heidelberg Engineering GmbH, Code F (Financial Support), Optovue Inc., Code F (Financial Support), Topcon Medical Systems Inc., Code F (Financial Support), Zeiss Meditec, Code P (Patent); Tara Javidi None; Robert Weinreb Aerie Pharmaceuticals, Code C (Consultant/Contractor), Allergan, Code C (Consultant/Contractor), Eyenovia, Code C (Consultant/Contractor), Heidelberg Engineering, Code F (Financial Support), Carl Zeiss Meditec, Code F (Financial Support), Konan Medical, Code F (Financial Support), Optovue, Code F (Financial Support), Centervue, Code F (Financial Support)
  • Footnotes
    Support  National Eye Institute R01EY029058, R01EY11008, R01EY19869, R01EY027510, R01EY026574, EY018926, P30EY022589, K99 EY030942, Unrestricted grants from Research to Prevent Blindness (New York, NY), Tobacco-Related Disease Research Program T31IP1511
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 832. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Alireza Kamalipour, Sasan Moghimi, Pooya Khosravi, Vahid Mohammadzadeh, Takashi Nishida, Eleonora Micheletti, Jo-Hsuan Wu, Golnoush Mahmoudinezhad, Mark Christopher, Linda Zangwill, Tara Javidi, Robert N Weinreb; Combining OCT and OCT-Angiography Longitudinal Measurements for the Prediction of Visual Field Progression in Glaucoma with Artificial Intelligence. Invest. Ophthalmol. Vis. Sci. 2022;63(7):832.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : To use longitudinal OCT and OCT-Angiography (OCTA) measurements to predict visual field (VF) progression in glaucoma suspect and glaucoma patients with a supervised machine learning approach.

Methods : 110 eyes of glaucoma suspect (33.6%) and glaucoma (66.4%) patients with a minimum of five 24-2 VF tests over an average follow-up duration of 4.1 years were included. Participants were required to have at least three pairs of optic nerve head and macula OCTA images during follow-up. VF progression was defined based on a composite measure including either a “likely progression event” on Guided Progression Analysis, a statistically significant negative slope of VF mean deviation, or a positive pointwise linear regression event. Gradient Boosting Classifier was used to predict the probability of VF progression based on different subsets of baseline and longitudinal OCT and OCTA input features at the global and regional levels. Areas-under-ROC curves (AUROC) were used to compare the classification accuracy of different models.

Results : VF progression was detected in 28 eyes (25.5%). The model that used combined baseline and longitudinal OCT and OCTA features at the global and regional level had the best classification accuracy for the prediction of VF progression (AUROC = 0.89 [95% CI: 0.82, 0.95]). Models including combined OCT and OCTA features had higher classification accuracy compared to those with individual subsets of OCT or OCTA features alone. Moreover, including regional measurements significantly improved the classification accuracy of the models compared to using global measurements alone. The addition of longitudinal rates of change of OCT and OCTA measurements as input features (AUROCs = 0.80-0.89) considerably increased the classification accuracy of the models with baseline measurements alone (AUROCs = 0.60 0.63). (Figure 1, all P-values for pairwise comparisons < 0.05)

Conclusions : Artificial intelligence techniques combining longitudinal OCT and OCTA measurements can predict clinically-relevant glaucomatous VF progression. Longitudinal OCTA measurements complement OCT-derived structural metrics for the prediction of functional VF loss in glaucoma patients.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

Figure 1.Areas-under-ROC curves (AUROC) plots of different models for the prediction of visual field progression.

Figure 1.Areas-under-ROC curves (AUROC) plots of different models for the prediction of visual field progression.

 

Figure 2.The relative influence of input features for the best model.

Figure 2.The relative influence of input features for the best model.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×