Related Experiment Video
Updated: Jul 7, 2025
05:52
Author Spotlight: Efficient Retinal Ganglion Cell Counting in Mouse Models of Glaucoma for Treatment Evaluation
Published on: October 4, 2024
1.0K
Deep learning framework for automated goblet cell density analysis in in-vivo rabbit conjunctiva
Seunghyun Jang1, Seonghan Kim2, Jungbin Lee2
1Department of Biomedical Engineering, Yonsei University, 1 Yonseidae-gil, Wonju-si, Gangwon-do, 26493, Republic of Korea.
Scientific Reports
|December 21, 2023
Summary
A new deep learning framework, DCAU-Net, accurately analyzes goblet cells (GCs) from eye images. This method aids in diagnosing ocular surface diseases by quantifying GC density and size.
Area of Science:
- Ophthalmology
- Biomedical Imaging
- Artificial Intelligence
Background:
- Goblet cells (GCs) are crucial for ocular surface health, producing mucins for tear film and immune tolerance.
- GC loss is a marker for various ocular surface diseases, necessitating precise diagnostic methods.
- Moxifloxacin-based fluorescence microscopy (MBFM) offers non-invasive, high-contrast visualization of GCs.
Purpose of the Study:
- To develop a robust deep learning framework for analyzing GC images obtained via MBFM.
- To enable quantitative assessment of GC density and morphology for diagnostic purposes.
Main Methods:
- Development of a dual-channel attention U-Net (DCAU-Net) deep learning model.
- Utilizing dual-channel convolution for texture and morphological feature extraction.
- Incorporating a global channel attention module for enhanced analysis.
- Validation using both normal and ocular surface damage rabbit models.
Main Results:
- DCAU-Net achieved 93.1% accuracy in GC segmentation and 94.3% in GC density estimation.
- Demonstrated spatial variations in GC density and size in normal rabbits.
- Showed decreased GC density and size in rabbits with ocular surface damage during recovery.
- GC analysis results correlated with histological findings.
Conclusions:
- DCAU-Net provides a robust and accurate method for analyzing GC images.
- This deep learning approach, combined with MBFM, can offer valuable information for diagnosing ocular surface diseases.
- The findings support the potential clinical application of DCAU-Net in precision ophthalmology.
