The STAGE Challenge is an international ophthalmology competition held by HDMI Lab of South China University of Technology and Sun Yat-sen Ophthalmic Center, Sun Yat-sen University at the MICCAI2023. MICCAI is a comprehensive academic conference in the fields of medical image computing and computer assisted intervention, and is the top conference in these fields. At the same time, organizers will organize the 10th OMIA workshop at MICCAI 2023.
This challenge uses Optical Coherence Tomography (OCT) images centered on the fovea to predict the results of the Visual Field (VF) test. OCT is the most widely used imaging method in ophthalmic examination. It provides objective cross-sectional information of the fundus structure and facilitates doctors to observe changes in the thickness of the structure, which is an important basis for the diagnosis of glaucoma [1]. Now, there is a lot of evidence supporting the role of OCT imaging in the detection of glaucoma [2-5]. VF test is a reference standard examination to assess visual function. This is a subjective test that requires the subject to remain calm and focused and cooperate with the doctor. Monocular perimetry takes about 15 minutes. It is the clinical standard to determine whether glaucoma optic nerve injury exists [6]. In contrast, a monocular OCT scan takes only about three seconds. In addition, there was a moderate to good correlation between retinal layer thickness and central field sensitivity or other markers of optic nerve function [7]. Therefore, this challenge focuses on how to use objective and easily accessible OCT images of structures to predict functional VF information. Based on this research, three VF information prediction tasks are proposed:
- Prediction of Mean Deviation (MD);
- Sensitivity map prediction;
- Pattern deviation probability map prediction.
The Sun Yat-sen Eye Center at Sun Yat-sen University will provide 400 OCT data and corresponding VF test reports with MD values, sensitivity maps and pattern deviation probability map labels. Among them, 200 macular OCT images and corresponding labels will be distributed to participating teams for model training. 100 macular OCT data will also be released in the preliminary round, and the evaluation platform will be opened. The remaining 100 macular OCT data will be released during the finals. From a technical point of view, the challenge involves computer vision research, where tasks 1 and 2 involve parametric regression problems, and task 3 involves classification problems. These studies are essential in computer-aided clinical diagnosis. From a biomedical perspective, the challenge seeks a mapping relationship between fundus structure and visual function, which is critical to understanding the underlying causes of visual deficits.
Please visit the URL of the challenge: