OCT Image Segmentation Model
This pretrained model performs automated segmentation of Optical Coherence Tomography (OCT) images using a U-Net–based deep learning architecture. It is designed to accurately delineate retinal structures and pathological regions from OCT B-scans, enabling quantitative retinal analysis and supporting ophthalmic research workflows.
Training Data
The model was trained on a private OCT dataset introduced in the following publication:
Li, J., Jin, P., Zhu, J., Zou, H., Xu, X., Tang, M., Zhou, M., Gan, Y., He, J., Ling, Y., et al. (2021). Multi-scale GCN-assisted two-stage network for joint segmentation of retinal layers and discs in peripapillary OCT images. Biomedical Optics Express, 12(4), 2204–2220.
1@article{li2021multi,
2 title={Multi-scale GCN-assisted two-stage network for joint segmentation of retinal layers and discs in peripapillary OCT images},
3 author={Li, Jiaxuan and Jin, Peiyao and Zhu, Jianfeng and Zou, Haidong and Xu, Xun and Tang, Min and Zhou, Minwen and Gan, Yu and He, Jiangnan and Ling, Yuye and others},
4 journal={Biomedical Optics Express},
5 volume={12},
6 number={4},
7 pages={2204--2220},
8 year={2021},
9 publisher={Optical Society of America}
10}
Dataset Availability
The training dataset is not publicly available. Researchers interested in accessing the data should contact the authors of the original publication directly and comply with any applicable data-sharing agreements and ethical requirements.
Intended Use
This model is intended for:
- Retinal layer and structure segmentation in OCT images
- Ophthalmic image analysis research
- Development and benchmarking of medical image segmentation methods
- Educational and experimental applications in deep learning for medical imaging
license: cc-by-2.0
