Lung Vessel Segmentation

该数据集来自VESSEL12挑战，其目的是比较从健康和患病人群的胸部计算机断层扫描中对肺血管进行（半）自动分割的方法。扫描来自各种来源，代表了各种临床上常见的扫描仪和协议。选择扫描，使得在大约一半的扫描中使用造影剂。大约一半的扫描包含肺气肿，结节或肺栓塞等异常情况。当前的最大切片间距为1毫米，大多数扫描（各向同性）。

This dataset is sourced from the VESSEL12 Challenge, whose core purpose is to compare (semi-)automatic lung vessel segmentation methods applied to chest computed tomography (CT) scans from both healthy and diseased patient cohorts. The scans are collected from diverse origins, representing a spectrum of clinically prevalent scanners and imaging protocols. Scans were selected to ensure that approximately half of the dataset was acquired with contrast medium administration. Roughly 50% of the included scans feature pathological abnormalities such as emphysema, pulmonary nodules, or pulmonary embolism. The maximum slice spacing across all scans is 1 mm, with the majority of scans being isotropic.

搜集汇总

背景与挑战

背景概述

该数据集来自VESSEL12挑战，包含23个胸部CT扫描及其血管分割标签，用于比较肺血管的（半）自动分割方法。扫描来自多样化的临床来源，约一半使用造影剂，一半包含肺气肿等异常，适用于医学图像分析研究。