Related Experiment Video
Updated: Aug 19, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
Choroidal layer segmentation in OCT images by a boundary enhancement network
Wenjun Wu1,2, Yan Gong3, Huaying Hao1
1Ningbo Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, China.
Frontiers in Cell and Developmental Biology
|November 28, 2022
Summary
This study introduces a new deep learning method for segmenting the choroid layer in Optical Coherence Tomography (OCT) images, improving accuracy for ophthalmic disease research and high myopia analysis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Choroidal morphological changes are linked to various eye diseases.
- Optical Coherence Tomography (OCT) provides high-resolution imaging of ocular tissues.
- Accurate choroid segmentation in OCT is challenging due to ambiguous boundaries with the sclera.
Purpose of the Study:
- To develop an advanced deep learning model for precise choroid segmentation in retinal OCT images.
- To address the ambiguity of the choroid-sclera interface in OCT scans.
- To enable detailed analysis of choroidal structure in ophthalmic conditions like high myopia.
Main Methods:
- A novel boundary-enhanced encoder-decoder architecture with a Boundary Enhancement Module (BEM) was proposed.
- The BEM incorporates Feature Extraction, Channel Enhancement, and Boundary Activation Branches.
- Soft key point maps were integrated to guide precise choroidal boundary segmentation.
Main Results:
- The proposed method demonstrated superior choroid segmentation performance compared to existing deep learning approaches.
- Qualitative and quantitative evaluations on three OCT datasets confirmed the method's effectiveness.
- The segmentation results facilitated the extraction of 2D and 3D features for analyzing normal and myopic subjects.
Conclusions:
- The developed boundary-enhanced network significantly improves choroid segmentation accuracy in OCT images.
- This technique aids in understanding the pathological mechanisms of ophthalmic diseases, particularly high myopia.
- The open-source code facilitates further research and clinical application.
