Related Experiment Video
Updated: May 24, 2025
07:12
Development of a Gaze-Contingent Display Framework Designed for Perceptual and Oculomotor Research with Simulated Central Vision Loss
Published on: April 11, 2025
260
An Intelligent Grading Model for Myopic Maculopathy Based on Long-Tailed Learning
Bo Zheng1,2, Chen Wang1, Maotao Zhang1
1School of Information Engineering, Huzhou University, Huzhou, China.
Translational Vision Science & Technology
|March 6, 2025
Summary
This study introduces an intelligent grading model for myopic maculopathy using long-tail learning and an improved LTBSoftmax loss function. The model enhances diagnostic accuracy and efficiency, offering a practical tool for clinicians.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Myopic maculopathy presents a long-tail data distribution challenge in automated grading.
- Existing models may struggle with data imbalance, impacting diagnostic accuracy and efficiency.
Purpose of the Study:
- To develop an intelligent grading model for myopic maculopathy using a long-tail learning framework.
- To address data imbalance with an improved LTBSoftmax loss function for enhanced grading capability and efficiency.
Main Methods:
- A dataset of 7529 color fundus photographs was utilized, with expert ophthalmologist annotations for ground truth.
- An intelligent grading model was constructed incorporating the LTBSoftmax loss function and ND Block for enhanced feature extraction.
- The model's performance was evaluated using standard grading metrics.
Main Results:
- The LTBSoftmax model achieved a κ coefficient of 88.89% in diagnosing four types of myopic maculopathy.
- The model demonstrates superior performance with high agreement with expert diagnoses.
- The model's compact size (18.7 MB) indicates high efficiency in storage and computational resources.
Conclusions:
- The intelligent grading system effectively improves myopic maculopathy classification using long-tailed learning strategies.
- This model serves as a practical grading tool for clinicians, especially in resource-limited settings.
