RETA
收藏资源简介:
RETA数据集是由浙江大学计算机科学与技术学院创建的,专注于视网膜血管树分析。该数据集包含81张图像,源自IDRiD数据集的第一个子集,并附有像素级别的血管掩模。创建过程中,采用了半自动化的粗到细工作流程进行血管像素的标注,并严格控制了标注者间的变异性和标注者内的变异性。数据集不仅包含二值血管掩模,还提供了动脉/静脉掩模、血管骨架、分支点、树和异常等详细标注。RETA数据集的应用领域广泛,包括自动血管分割算法的开发与评估,以及跨模态管状结构分割的研究,旨在解决视网膜血管分析中的通用性问题。
The RETA dataset was developed by the College of Computer Science and Technology, Zhejiang University, and focuses on retinal vascular tree analysis. It contains 81 images sourced from the first subset of the IDRiD dataset, paired with pixel-level vascular masks. During the dataset construction, a semi-automated coarse-to-fine workflow was utilized for annotating vascular pixels, with strict control over inter-annotator and intra-annotator variability. The dataset not only includes binary vascular masks but also provides detailed annotations such as artery/vein masks, vascular skeletons, bifurcation points, vascular trees, and abnormalities. The RETA dataset has broad application scenarios, including the development and evaluation of automated vascular segmentation algorithms, as well as research on cross-modal tubular structure segmentation, aiming to solve the generalizability challenges in retinal vascular analysis.

- 1The RETA Benchmark for Retinal Vascular Tree Analysis浙江大学计算机科学与技术学院 · 2021年



