DRIVE(Digital Retinal Images for Vessel Extraction)

用于血管提取的数字视网膜图像 (DRIVE) 数据集是用于视网膜血管分割的数据集。它由总共 40 张 JPEG 彩色眼底图像组成；其中病理异常7例。这些图像来自荷兰的糖尿病视网膜病变筛查项目。图像是使用佳能 CR5 非散瞳 3CCD 相机获得的，FOV 等于 45 度。每个图像分辨率为 584*565 像素，每个颜色通道（3 个通道）有 8 位。这组 40 幅图像平均分为 20 幅图像作为训练集和 20 幅图像作为测试集。在这两个集合中，对于每个图像，都有一个直径约为 540 像素的圆形视场 (FOV) 掩模。在训练集中，对于每张图像，已经应用了由眼科专家进行的一次手动分割。在测试集中，对于每个图像，两个不同的观察者应用了两个手动分割，其中第一个观察者分割被接受为性能评估的基本事实。

The Digital Retinal Images for Vessel Extraction (DRIVE) dataset is a dataset designed for retinal blood vessel segmentation. It comprises a total of 40 JPEG color fundus images, 7 of which exhibit retinal pathological abnormalities. These images were collected from a diabetic retinopathy screening program in the Netherlands. All images were acquired using a Canon CR5 non-mydriatic 3CCD camera, with a field of view (FOV) of 45 degrees. Each image has a resolution of 584 × 565 pixels, with 8-bit depth for each of the three color channels. The 40 images are evenly split into a training set of 20 images and a test set of 20 images. For each image in both sets, there is a circular field-of-view (FOV) mask with a diameter of approximately 540 pixels. For each image in the training set, one manual segmentation annotation performed by an ophthalmologist is provided. For each image in the test set, two manual segmentations produced by two independent observers are available, with the segmentation from the first observer serving as the ground truth for performance evaluation.

搜集汇总

背景与挑战

背景概述

DRIVE数据集是一个用于视网膜血管分割的医疗图像数据集，包含40张彩色眼底图像，其中7张有病理异常，图像分辨率为584*565像素。数据集分为训练集和测试集各20张，提供手动分割标注，主要用于评估血管分割算法，发布于2004年，适用于医学图像分析研究。