RITE (Retinal Images vessel Tree Extraction)

RITE（视网膜图像血管树提取）是一个数据库，可以对视网膜眼底图像上的动脉和静脉的分割或分类进行比较研究，该数据库是基于公共可用的 DRIVE 数据库（用于血管提取的数字视网膜图像）建立的。 RITE 包含 40 组图像，平均分为训练子集和测试子集，与 DRIVE 相同。这两个子集是从 DRIVE 中相应的两个子集构建的。每套都有眼底照片、血管参考标准和动脉/静脉 (A/V) 参考标准。眼底照片继承自 DRIVE。对于训练集，船舶参考标准是 DRIVE 的 1st_manual 的修改版本。对于测试集，容器参考标准是来自 DRIVE 的 2nd_manual。对于 A/V 参考标准，四种类型的容器根据容器参考标准使用四种颜色进行标记。动脉标记为红色；静脉标记为蓝色；动脉和静脉的重叠用绿色标记；不确定的血管用白色标记。眼底照片为 tif 格式。容器参考标准和A/V参考标准为png格式。该数据集在我们的论文中进行了更详细的描述，如果您以任何方式使用该数据集，您都会引用该数据集：Hu Q、Abràmoff MD、Garvin MK。自动分离低对比度彩色视网膜图像中的二元重叠树。 Med Image Comput Comput Assist 访谈。 2013；16（第 2 部分）：436-43。 PubMed PMID：24579170 https://doi.org/10.1007/978-3-642-40763-5_54

RITE (Retinal Image Vessel Tree Extraction) is a database designed for comparative studies on the segmentation or classification of arteries and veins in retinal fundus images, which is built upon the publicly available DRIVE (Digital Retinal Images for Vessel Extraction) database. RITE comprises 40 image sets, evenly split into training and test subsets following the same partition as the DRIVE database. These two subsets are constructed from the corresponding subsets of DRIVE. Each set includes a fundus photograph, a vessel reference standard, and an artery/vein (A/V) reference standard. The fundus photographs are sourced from the DRIVE database. For the training subset, the vessel reference standard is a modified version of the 1st_manual from DRIVE, while for the test subset, the vessel reference standard is the 2nd_manual from DRIVE. For the A/V reference standard, four categories of vessels are labeled with four distinct colors based on the vessel reference standard: arteries are marked red, veins are marked blue, overlapping regions of arteries and veins are marked green, and uncertain vessels are marked white. All fundus photographs are in TIFF format, whereas both the vessel reference standards and A/V reference standards are in PNG format. This dataset is described in greater detail in our accompanying paper. Please cite this dataset if you use it in any capacity: Hu Q, Abràmoff MD, Garvin MK. Automatic separation of binary overlapping trees in low-contrast color retinal images. Med Image Comput Comput Assist Interv. 2013;16(Part 2):436-443. PubMed PMID: 24579170 https://doi.org/10.1007/978-3-642-40763-5_54

搜集汇总

背景与挑战

背景概述

RITE是一个用于视网膜图像血管树提取的医学图像分割数据集，基于DRIVE数据库构建，包含40组图像（训练和测试子集各20组），每套提供眼底照片、血管参考标准及动脉/静脉标记参考标准。该数据集主要用于动脉和静脉的分割或分类比较研究，支持医疗CV领域的算法开发与评估，由爱荷华大学于2013年发布。