June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Predicting Visual Outcome in Patients with Idiopathic Epiretinal Membrane Using A Novel Convolutional Neural Network
Author Affiliations & Notes
  • Tsai-Chu Yeh
    Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan
  • Yu-Chieh Ko
    Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan
  • Yu-Bai Chou
    Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan
  • Shih-Jen Chen
    Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan
  • An-Chun Luo
    Industrial technology research institute, Taiwan
  • Yu-Shan Deng
    Industrial technology research institute, Taiwan
  • Footnotes
    Commercial Relationships   Tsai-Chu Yeh None; Yu-Chieh Ko None; Yu-Bai Chou None; Shih-Jen Chen None; An-Chun Luo None; Yu-Shan Deng None
  • Footnotes
    Support   None
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 2083 – F0072. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Tsai-Chu Yeh, Yu-Chieh Ko, Yu-Bai Chou, Shih-Jen Chen, An-Chun Luo, Yu-Shan Deng; Predicting Visual Outcome in Patients with Idiopathic Epiretinal Membrane Using A Novel Convolutional Neural Network. Invest. Ophthalmol. Vis. Sci. 2022;63(7):2083 – F0072.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Vitrectomy with epiretinal membrane (ERM) peeling is considered the current standard treatment for ERM; however, some patients may experience unfavorable prognosis with limited visual improvement. It remains as a challenge to determine a more rational surgical indication and timing but the guidance or predictive model is lacking currently. Thus, we aim to develop a deep convolutional neural network (CNN) that enables prediction of postoperative visual outcomes based on preoperative optical coherence tomography (OCT) images and clinical prognostic factors to refine surgical decision-making.

Methods : A total of 529 patients with idiopathic ERM who received standard vitrectomy with ERM peeling surgery between January 1, 2014 and June 1, 2020 were enrolled. The newly developed CNN model was introduced to predict postoperative visual acuity (VA) outcome (improvement ≥ 2 lines in Snellen chart or not) 12th months after surgery based on preoperative cross section OCT images and clinical factors, including age, gender and preoperative VA.

Results : The model demonstrated an overall accuracy for visual outcome prediction of 90.2% (95% CI, 79.0%-95.7%) with an AUC of 97.8% (95% CI, 86.8%-98.0%), sensitivity of 87.0% (95% CI, 67.9%-95.5%), specificity of 92.9% (95% CI, 77.4%-98.0%), precision of 0.909, recall of 0.870 and F1 score of 0.889. The heatmaps identified the critical area for prediction as the fovea subjected to tangential traction of the proliferative membrane and the adjacent ellipsoid zone of photoreceptors.

Conclusions : The novel CNN model demonstrated high accuracy in automated prediction of visual outcome based on preoperative macular OCT images. Our study suggests that deep learning has the potential to weigh and leverage the complete data of the patient, and simultaneously process a broad range of clinical information including OCT images. The model has the potential to predict visual outcome of ERM surgery through weighing and leveraging clinical information including OCT images. This approach may be helpful in setting personalized therapeutic strategy for ERM management.

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

×
×

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.

×