Abstract: Accurate segmentation of the optic disc and optic cup is crucial for the detection of glaucoma. Deep learning technology has achieved remarkable results in the field of optic disc and cup segmentation, significantly improving segmentation accuracy. Starting from the current state of research on deep learning technology for optic disc and cup segmentation, this summary includes the commonly used datasets for segmentation, detailing their content, purpose, and acquisition paths, as well as key metrics for evaluating segmentation and model performance. An analysis of four primary research methods in optic disc and cup segmentation is presented： methods based on multi-scale, fusion of attention mechanisms, adversarial learning mechanisms, and ensemble learning methods. The advantages and disadvantages of these methods are discussed, and their performance metrics on commonly used public datasets are summarized. Finally, challenges faced by optic disc and cup segmentation in glaucoma detection are explored, and future research directions are outlined, with the aim of providing a reference for further studies in this field.

Key words: glaucoma, image segmentation, optic disc, optic cup

摘要： 精准的视盘与视杯分割对于青光眼的检测至关重要。近年来，深度学习技术在视盘与视杯分割领域取得了优异的成果，显著提升了分割精度。从深度学习技术在视盘与视杯分割的研究现状出发，归纳了视盘与视杯分割的常用数据集，包括其内容、用途和获取路径；概述了评估分割性能与模型性能的关键指标。分析了视盘与视杯分割中四类主要研究方法：基于多尺度的方法、注意力机制的融合、对抗学习机制及集成学习方法。对这些方法进行了优缺点分析，总结了它们在常用公开数据集上的性能指标。最后，探讨了视盘与视杯分割在青光眼检测中所面临的挑战，并展望了未来的研究方向，旨在为该领域的进一步研究提供参考。

关键词: 青光眼, 图像分割, 视盘, 视杯