Related Experiment Video
Updated: Sep 16, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Deep learning for diagnosing and grading pterygium: A systematic review and meta-analysis
Ethan W W Tiong1, Carine Y S Soon1, Zun Zheng Ong2
1, School of Medical Sciences, University of Manchester, UK.
Computers in Biology and Medicine
|July 11, 2025
Summary
Deep learning (DL) models show high accuracy in detecting and grading pterygium, potentially matching expert ophthalmologists. Further research is needed due to methodological limitations.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Deep learning (DL) shows promise in diagnosing eye conditions.
- Pterygium diagnosis and grading accuracy using DL requires further investigation.
Purpose of the Study:
- To systematically review and meta-analyze the diagnostic accuracy of DL models for pterygium detection and severity assessment.
- To compare DL model performance against clinical experts.
Main Methods:
- Systematic search of EMBASE, MEDLINE, and clinical registries (1974-Feb 2025).
- Inclusion of peer-reviewed studies on AI algorithms for pterygium diagnosis/grading.
- Risk of bias assessment using QUADAS-2 and bivariate random-effects models for diagnostic accuracy analysis.
Main Results:
- 20 studies (45,913 images) were included; no external validation data.
- Summary sensitivity/specificity for pterygium diagnosis: 98.1% (96.4-99.1) and 99.1% (98.0-99.6).
- Summary sensitivity/specificity for severity grading: 91.2% (87.7-93.7) and 92.9% (88.4-95.8).
Conclusions:
- DL models demonstrate high accuracy for pterygium diagnosis and grading, comparable to clinical experts.
- Methodological limitations (lack of external validation, study designs) necessitate cautious interpretation.
- Future studies require transparent reporting, prospective designs, and external validation for clinical translation.
