CAMUS(Cardiac Acquisitions for Multi-structure Ultrasound Segmentation)

该项目的目标是向社区提供所有材料，以解决二维超声序列（两腔和四腔视图）的超声心动图图像分割和体积估计问题。为此，以下解决方案设置了用于 2D 超声心动图评估（据我们所知）的最大的公开可用且完全注释的数据集的介绍。 CAMUS 数据集包含从 500 名患者获得的二维心尖四腔和两腔视图序列，可用于下载部署专用 Girder 在线平台。该平台旨在以可重复的方式评估心脏结构（左心室心内膜和心外膜以及左心房边界）分割方法的性能以及临床指标（左心室容积和射血分数）的提取。 CAMUS 在线平台现已可用，只要数据与临床研究相关，就会保持并保持开放。

The goal of this project is to provide the community with all necessary materials to address the tasks of echocardiographic image segmentation and volume estimation for 2D ultrasound sequences (apical two-chamber and apical four-chamber views). To this end, the proposed solution introduces the largest publicly available and fully annotated dataset for 2D echocardiographic evaluation, to the best of our knowledge. The CAMUS dataset comprises 2D apical four-chamber and two-chamber view sequences acquired from 500 patients, which can be downloaded via the dedicated Girder online platform. This platform is designed to evaluate the performance of cardiac structure segmentation methods (including left ventricular endocardium, epicardium, and left atrial boundary) and the extraction of clinical metrics such as left ventricular volume and ejection fraction in a reproducible manner. The CAMUS online platform is now publicly available, and it will remain open and continuously maintained as long as the data remains relevant to clinical research.

搜集汇总

背景与挑战

背景概述

CAMUS数据集是一个公开的医疗心脏超声图像分割数据集，包含500名患者的二维心尖四腔和两腔视图序列，用于评估心脏结构分割和临床指标提取。该数据集由挪威科技大学和里昂大学于2019年发布，通过在线平台支持可重复的性能评估。