Related Experiment Video
Updated: Jul 26, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.8K
JOINT MOTION CORRECTION AND 3D SEGMENTATION WITH GRAPH-ASSISTED NEURAL NETWORKS FOR RETINAL OCT
Yiqian Wang1, Carlo Galang2, William R Freeman2
1Department of Electrical and Computer Engineering, University of California, San Diego.
Proceedings. International Conference on Image Processing
|June 21, 2023
Summary
This study introduces neural networks to simultaneously correct eye motion and segment retinal layers in 3D Optical Coherence Tomography (OCT) imaging. This approach improves consistency and accuracy in retinal layer segmentation for ophthalmology applications.
Area of Science:
- Ophthalmology
- Medical Imaging
- Biomedical Engineering
Background:
- Optical Coherence Tomography (OCT) is a key non-invasive 3D imaging modality in ophthalmology.
- Accurate retinal layer segmentation is crucial for OCT-Angiography and disease diagnosis.
- Involuntary eye movements cause motion artifacts, degrading OCT image quality and segmentation accuracy.
Purpose of the Study:
- To develop a novel neural network approach for joint eye motion correction and retinal layer segmentation in 3D OCT data.
- To ensure consistency in segmentation across adjacent B-scans by integrating 3D OCT information.
- To enhance the reliability of OCT-based retinal analysis.
Main Methods:
- Proposed a deep learning framework utilizing 3D OCT data.
- Implemented a joint model for simultaneous motion artifact correction and retinal layer segmentation.
- Evaluated the method against conventional and 2D deep learning segmentation techniques.
Main Results:
- Demonstrated significant visual improvements in OCT images after motion correction and segmentation.
- Achieved superior quantitative results in retinal layer segmentation compared to existing 2D methods.
- Showcased enhanced consistency of segmentation across neighboring B-scans.
Conclusions:
- The proposed joint 3D OCT motion correction and segmentation method offers substantial advantages over 2D approaches.
- This technique improves the accuracy and reliability of retinal layer segmentation in the presence of eye motion.
- The findings support the clinical utility of advanced OCT image processing for ophthalmological applications.
