SegTHOR

该数据集专为2D分割任务设计，源自3D医疗影像，特别关注于胸腔器官。原始的3D数据沿着Z轴被分割成2D图像，并且在训练数据集中移除了纯黑色的地面真实值切片。该任务涉及的是2D CT图像分割。

This dataset is specifically designed for 2D segmentation tasks, derived from 3D medical images with a particular focus on thoracic organs. The original 3D data was split into 2D images along the Z-axis, and pure black ground truth slices were removed from the training dataset. This task involves 2D CT image segmentation.

搜集汇总

背景与挑战

背景概述

SegTHOR数据集是一个用于CT图像中风险器官自动分割的挑战数据集，包含60个患者的CT扫描，分为40个训练样本和20个测试样本。数据集关注心脏、主动脉、气管和食道的分割，旨在提高放射治疗计划中器官分割的准确性和效率。