Related Experiment Video
Updated: Jul 17, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Adaptive contour prediction in postoperative OCT imaging using domain-adaptive generative adversarial networks
Shaopeng Liu1, Kai Wang2, Xiaohang Wu3
1Schoolof Computer Science, Guangdong Polytechnic Normal University, Guangzhou, 510665, China; Guangdong Key Laboratory of Big Data Analysis and Processing, Guangzhou, 510006, China.
Photodiagnosis and Photodynamic Therapy
|June 28, 2026
Summary
This study introduces a novel domain-adaptive generative adversarial network (DaCGAN) for predicting postoperative Optical Coherence Tomography (OCT) contours from preoperative images, improving diagnostic accuracy for eye diseases.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Optical Coherence Tomography (OCT) is crucial for diagnosing ophthalmic conditions.
- Predicting postoperative OCT contours from preoperative images is challenging due to limited data and incomplete structures.
- Existing methods struggle with domain adaptation and contour prediction accuracy.
Purpose of the Study:
- To develop a domain-adaptive generative adversarial network (DaCGAN) for accurate postoperative OCT contour prediction.
- To address data limitations and incomplete contour structures in OCT imaging.
- To improve the clinical utility of OCT in diagnosis and treatment planning.
Main Methods:
- Introduced a domain-adaptive generative adversarial network (DaCGAN).
- Integrated data from different ophthalmic diseases.
- Utilized preoperative contour information to enhance prediction accuracy.
- Validated the model on Diabetic Macular Edema and Retinal Vein Occlusion datasets.
Main Results:
- DaCGAN demonstrated superior performance compared to existing methods.
- Significant improvements observed in Structure Similarity Index (SSIM), Mean Squared Error (MSE), and Peak Signal-to-Noise Ratio (PSNR) metrics.
- The model effectively predicted postoperative OCT contours with enhanced accuracy.
Conclusions:
- DaCGAN offers a cost-effective solution for postoperative OCT image prediction.
- The proposed method shows potential for benefiting clinical diagnosis and treatment planning in ophthalmology.
- Domain adaptation and integration of preoperative data are key to improving OCT contour prediction.