胎儿常规产前筛查超声影像人工标注数据集

本数据集适用于以下场景： • 开发能解析医生可视化标注的AI算法（如结合OCR与目标检测的弱监督学习模型）； • 训练胎儿标准切面自动识别、器官定位与生物参数提取模型； • 构建基层医师产前超声培训与质控系统； • 支持多模态（图像+视频）融合分析的AI原型验证； • 用于医疗器械（如产科AI软件）的算法训练与临床性能评估。 主要面向具备医疗数据处理资质的医院、科研机构、高校及医疗器械企业。

This dataset is applicable to the following scenarios:
• Develop AI algorithms that parse physician visual annotations (e.g., weakly-supervised learning models combining OCR and object detection);
• Train models for automatic recognition of fetal standard ultrasound planes, organ localization and biometric parameter extraction;
• Build prenatal ultrasound training and quality control systems for primary care physicians;
• Support AI prototype validation for multimodal (image + video) fusion analysis;
• Conduct algorithm training and clinical performance evaluation for medical devices (e.g., obstetric AI software).

This dataset is primarily targeted at hospitals, research institutions, universities and medical device enterprises that possess valid qualifications for medical data processing.

搜集汇总

背景与挑战

背景概述

该数据集是一个专门针对孕中晚期胎儿常规产前筛查的超声影像人工标注数据集，包含来自真实临床检查的静态图像和动态视频，覆盖颅脑、心脏等多个关键解剖区域。所有数据均由具备资质的产科超声医师依据国际和国内临床指南进行可视化标注，标注形式包括轮廓、箭头、文本等，精准反映了实际诊疗实践。该数据集适用于开发AI辅助诊断模型、支持医师培训与质控系统以及进行医疗器械算法验证等多种医疗人工智能应用场景。