Related Experiment Video
Updated: Oct 9, 2025
07:23
Optimization of the Retinal Vein Occlusion Mouse Model to Limit Variability
Published on: August 6, 2021
2.9K
Artificial intelligence can assist with diagnosing retinal vein occlusion
Qiong Chen1, Wei-Hong Yu2, Song Lin1
1Tianjin Key Laboratory of Retinal Functions and Diseases, Tianjin Branch of National Clinical Research Center for Ocular Disease, Eye Institute and School of Optometry, Tianjin Medical University Eye Hospital, Tianjin 300384, China.
International Journal of Ophthalmology
|December 20, 2021
Summary
Deep learning (DL) models demonstrate high accuracy in identifying retinal vein occlusion (RVO) and its associated lesions from color fundus photographs (CFPs). These AI tools can aid in early RVO diagnosis and reduce vision impairment.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Retinal vein occlusion (RVO) poses a significant threat to vision, necessitating early detection and treatment.
- The increasing prevalence of RVO and a shortage of trained ophthalmologists create a demand for efficient screening tools.
Purpose of the Study:
- To develop and evaluate deep learning (DL) models for automated screening of RVO using color fundus photographs (CFPs).
- To assess the performance of DL models in both disease recognition and lesion segmentation for RVO.
Main Methods:
- Trained and validated multiple DL models on 8600 CFPs for disease recognition and lesion segmentation.
- Selected superior Inception-v3 (disease recognition) and DeepLab-v3 (lesion segmentation) models.
- Validated the selected models on an independent external test set of 224 CFPs from 130 patients.
Main Results:
- The Inception-v3 model achieved high accuracy in RVO recognition (sensitivity 0.93, specificity 0.99, F1 0.95, AUC 0.99).
- The DeepLab-v3 model demonstrated strong performance in segmenting RVO lesions (sensitivity 0.74, specificity 0.97, F1 0.83).
Conclusions:
- DL models show significant potential for accurate RVO detection and lesion identification from CFPs.
- These AI tools can assist ophthalmologists in early RVO diagnosis, potentially reducing vision loss.
