Related Experiment Video
Updated: Jun 21, 2025
08:20
Author Spotlight: AI-Driven Trypanosome Species Detection from Microscopic Images
Published on: October 27, 2023
1.4K
Deep learning for multi-type infectious keratitis diagnosis: A nationwide, cross-sectional, multicenter study.
Zhongwen Li1,2, He Xie2, Zhouqian Wang2
1Ningbo Key Laboratory of Medical Research on Blinding Eye Diseases, Ningbo Eye Institute, Ningbo Eye Hospital, Wenzhou Medical University, Ningbo, 315000, China.
NPJ Digital Medicine
|July 6, 2024
Summary
A new deep learning system, DeepIK, can accurately diagnose infectious keratitis from slit-lamp images. This AI tool aids ophthalmologists in rapid identification and treatment of bacterial, fungal, viral, and amebic keratitis.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Infectious keratitis is a leading cause of global corneal blindness.
- Prompt diagnosis is crucial for effective management but current methods like corneal scraping cultures are slow and often inaccurate.
Purpose of the Study:
- To develop and validate a deep learning system (DeepIK) for diagnosing infectious keratitis.
- To assess DeepIK's ability to differentiate between bacterial, fungal, viral, amebic, and noninfectious keratitis using slit-lamp images.
Main Methods:
- A deep learning system, DeepIK, was trained and tested on 23,055 slit-lamp images from 12 clinical centers.
- Performance was evaluated using internal, external, and prospective datasets.
- DeepIK's diagnostic accuracy was compared against three state-of-the-art algorithms.
Main Results:
- DeepIK achieved high diagnostic performance across all datasets (Area Under the ROC Curve > 0.96).
- The system outperformed DenseNet121, InceptionResNetV2, and Swin-Transformer algorithms.
- DeepIK demonstrated capability in identifying and differentiating various types of keratitis.
Conclusions:
- DeepIK shows significant potential as a tool to assist ophthalmologists in the rapid and accurate diagnosis of infectious keratitis.
- This AI-driven approach can facilitate timely and targeted treatment, improving patient outcomes.
