Comprehensive Liver CT Landmark Pair Dataset
收藏资源简介:
本研究介绍了首个大规模肝脏CT变形图像注册(DIR)地标数据集,由杜克大学创建。该数据集包含30对增强CT图像,总计2028个地标点,平均每对图像约68个地标点,主要分布在肝脏血管分叉处。数据集通过半自动处理流程生成,包括自动分割肝脏和血管、自动检测地标点,并通过手动验证确保准确性。该数据集旨在为DIR算法提供密集的基准,用于定量评估肝脏CT图像的注册精度,适用于手术规划、放射治疗和热消融等临床应用。
This study introduces the first large-scale liver CT deformable image registration (DIR) landmark dataset, created by Duke University. This dataset consists of 30 pairs of contrast-enhanced CT images, with a total of 2028 landmark points, averaging approximately 68 landmarks per image pair, which are mainly distributed at the bifurcations of hepatic blood vessels. The dataset is generated via a semi-automated processing pipeline, which includes automatic segmentation of the liver and blood vessels, automatic detection of landmark points, and manual verification to ensure accuracy. This dataset aims to provide a dense benchmark for DIR algorithms, used to quantitatively evaluate the registration accuracy of liver CT images, and is applicable to clinical applications such as surgical planning, radiation therapy, and thermal ablation.

- 1A comprehensive liver CT landmark pair dataset for evaluating deformable image registration algorithms杜克大学 · 2024年



