Related Experiment Video
Updated: Sep 17, 2025
07:22
Quantitative Fundus Autofluorescence for the Evaluation of Retinal Diseases
Published on: March 11, 2016
11.5K
Generative artificial intelligence for fundus fluorescein angiography interpretation and human expert evaluation
An Shao1, Xiaocong Liu2, Wenyue Shen1
1Zhejiang University, Eye Center of Second Affiliated Hospital, School of Medicine, China. Zhejiang Provincial Key Laboratory of Ophthalmology. Zhejiang Provincial Clinical Research Center for Eye Diseases. Zhejiang Provincial Engineering Institute on Eye Diseases, Hangzhou, China.
NPJ Digital Medicine
|July 2, 2025
Summary
InterpreFFA, an AI tool, aids ophthalmologists in interpreting fundus fluorescein angiography (FFA) reports. It significantly boosts diagnostic accuracy and reduces reporting time for chorioretinal diseases.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Fundus fluorescein angiography (FFA) is crucial for diagnosing chorioretinal diseases but demands expert interpretation.
- Current automated FFA interpretation lacks robust models and evaluation metrics.
Purpose of the Study:
- To introduce InterpreFFA, a novel AI framework for automated FFA report generation.
- To emulate ophthalmologists' diagnostic decision-making process in FFA interpretation.
Main Methods:
- Developed a diagnosis-supervised contrastive learning framework (InterpFFA).
- Validated on multi-center datasets, comparing against baseline models.
- Conducted a simulated clinical trial with ophthalmology residents and board-certified ophthalmologists.
Main Results:
- InterpFFA demonstrated superior performance and generalization over baseline models.
- Significantly improved diagnostic accuracy (85.55% to 90.34%, p<0.05) and reduced reporting time (153.93s to 108.08s, p<0.001).
- AI-generated reports received slightly lower scores than manual reports (4.12 vs. 4.38, p<0.01).
Conclusions:
- InterpFFA shows promise as a cost-effective tool to enhance clinical efficiency in FFA interpretation.
- The AI framework can assist ophthalmologists, improving diagnostic accuracy and workflow speed.
- Further development may bridge the quality gap between AI-generated and manual reports.
