The retinal vasculature provides important clues in the diagnosis and monitoring of systemic diseases including hypertension and diabetes. The microvascular system is of primary involvement in such conditions, and the retina is the only anatomical site where the microvasculature can be directly observed. The objective assessment of retinal vessels has long been considered a surrogate biomarker for systemic vascular diseases, and with recent advancements in retinal imaging and computer vision technologies, this topic has become the subject of renewed attention. In this paper, we present a novel dataset, dubbed RAVIR, for the semantic segmentation of Retinal Arteries and Veins in Infrared Reflectance (IR) imaging. It enables the creation of deep learning-based models that distinguish extracted vessel type without extensive post-processing. We propose a novel deep learning-based methodology, denoted as SegRAVIR, for the semantic segmentation of retinal arteries and veins and the quantitative measurement of the widths of segmented vessels. Our extensive experiments validate the effectiveness of SegRAVIR and demonstrate its superior performance in comparison to stateof-the-art models. Additionally, we propose a knowledge distillation framework for the domain adaptation of RAVIR pretrained networks on color images. We demonstrate that our pretraining procedure yields new state-of-the-art benchmarks on the DRIVE, STARE, and CHASE DB1 datasets.
- RAVIR dataset can be downloaded from this link. By downloading RAVIR dataset, the user agrees to the data usage protocols (see below).
- Test set evaluation and leaderboard are now available in the challenge website.
- [Jul 2022] RAVIR Dataset is now publicly available !
- [Mar 2022] The RAVIR paper has been accepted to IEEE Journal of Biomedical Health Informatics.
The images in RAVIR dataset were captured using infrared (815nm) Scanning Laser Ophthalmoscopy (SLO), which in addition to having higher quality and contrast, is less affected by opacities in optical media and pupil size. RAVIR images are sized at 768 × 768, captured using a Heidelberg Spectralis camera with a 30° FOV and compressed in the Portable Network Graphics (PNG) format. Each pixel in the images has a reference length of 12.5 microns. RAVIR project was carried out with the approval of the Institutional Review Board at UCLA and adhered to the tenets of the Declaration of Helsinki.
RAVIR dataset is distributed under Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) Licence and may only be used for non-commercial purposes. The following manuscripts must be cited if RAVIR dataset is used in any instances:
[1]: Hatamizadeh, A., Hosseini, H., Patel, N., Choi, J., Pole, C., Hoeferlin, C., Schwartz, S. and Terzopoulos, D., 2022. RAVIR: A Dataset and Methodology for the Semantic Segmentation and Quantitative Analysis of Retinal Arteries and Veins in Infrared Reflectance Imaging. IEEE Journal of Biomedical and Health Informatics.
[2]: Hatamizadeh, Ali. An Artificial Intelligence Framework for the Automated Segmentation and
Quantitative Analysis of Retinal Vasculature. University of California, Los Angeles, 2020.
@article{hatamizadeh2022ravir,
title={RAVIR: A Dataset and Methodology for the Semantic Segmentation and Quantitative Analysis of Retinal Arteries and Veins
in Infrared Reflectance Imaging},
author={Hatamizadeh, Ali and Hosseini, Hamid and Patel, Niraj and Choi, Jinseo and Pole, Cameron and Hoeferlin, Cory and
Schwartz, Steven and Terzopoulos, Demetri},
journal={IEEE Journal of Biomedical and Health Informatics},
year={2022},
publisher={IEEE}
}
@book{hatamizadeh2020artificial,
title={An Artificial Intelligence Framework for the Automated Segmentation and Quantitative Analysis of Retinal Vasculature},
author={Hatamizadeh, Ali},
year={2020},
publisher={University of California, Los Angeles},
}