UK Biobank Fundus Retinal Images

该数据集包含了来自78,346位个体的右眼80,966张眼底视网膜图像，用于训练和评估一个专注于高血压疾病分类的模型。数据集还包括了丰富的元数据，如年龄、BMI、自我报告的饮酒量、种族和遗传信息。数据被分层并分为训练集、验证集和测试集，以便进行精确的模型训练和评估。该数据集的规模为78,346位个体的80,966张图像，任务是对视网膜图像进行高血压分类。This dataset contains 80,966 fundus retinal images of the right eyes from 78,346 individuals, which is utilized for training and evaluating a model dedicated to hypertension classification. The dataset also provides comprehensive metadata including age, BMI, self-reported alcohol consumption, race, and genetic information. The data is stratified and divided into training, validation, and test sets to facilitate precise model training and evaluation. Comprising 80,966 images from 78,346 individuals, this dataset focuses on the task of hypertension classification for retinal fundus images.

搜集汇总

背景与挑战

背景概述

该数据集基于英国生物银行（UK Biobank）的视网膜图像，专注于分析医学影像分类模型中的偏见问题。数据集包含右眼视网膜图像（数据字段21015），并整合了敏感属性（如采集中心），用于公平性基准测试和偏见缓解研究。由于数据使用协议限制，图像数据不公开共享，但提供了完整的预处理和实验代码框架。