Upper Ankle Joint Synthetic Dataset
收藏资源简介:
该数据集由吕贝克大学等机构创建,名为上踝关节合成数据集,包含3077对深度图像与X光图像,专为患者姿态评估任务设计。数据源自CT扫描,通过框架生成不同姿态与诊断质量的合成图像,涵盖多种体型模拟和病变增强。创建过程包括表面提取、点云增强、场景合成及蒙特卡洛模拟生成X光片,确保临床真实性。数据集旨在解决实际临床中获取配对深度图像与X光片的伦理与监管难题,用于训练AI模型自动评估患者姿态,提升放射诊断质量并减少重复曝光。
This dataset was created by institutions including the University of Lübeck, and is named the Superior Ankle Joint Synthetic Dataset. It contains 3077 pairs of depth images and X-ray images, specifically designed for patient posture assessment tasks. The data originates from CT scans, and synthetic images with varying postures and diagnostic qualities are generated via a framework, covering multiple body type simulations and lesion enhancements. Its creation process includes surface extraction, point cloud enhancement, scene synthesis, and Monte Carlo simulation for X-ray image generation, ensuring clinical authenticity. The dataset aims to address the ethical and regulatory challenges of acquiring paired depth images and X-ray images in actual clinical settings, and is used to train AI models for automatic patient posture assessment, thereby improving the quality of radiological diagnosis and reducing repeated radiation exposure.





