BraTS 2017

该数据集名为BraTS17，包含了280个3D脑部体积数据，并附有参考的分割结果，涵盖了高低级别胶质瘤的扫描。数据集包含了多种成像模式：原始T1加权、T1增强、T2加权以及T2液体抑制反转恢复（FLAIR）。此外，该数据集被划分为训练集（235例）、验证集（20例）和测试集（30例）。规模总计280例，任务是对脑肿瘤进行分割。

This dataset is named BraTS17. It contains 280 3D brain volumetric data paired with reference segmentation results, covering scans of both low-grade and high-grade gliomas. The dataset includes multiple imaging modalities: native T1-weighted, T1-weighted contrast-enhanced, T2-weighted, and T2 fluid-attenuated inversion recovery (FLAIR). Additionally, this dataset is split into a training set (235 cases), a validation set (20 cases) and a test set (30 cases), with a total of 280 cases overall. The task of this dataset is brain tumor segmentation.

搜集汇总

背景与挑战

背景概述

BraTS 2017数据集是一个多模态脑肿瘤分割挑战赛的数据集，包含多机构临床获取的术前多模态MRI扫描，涵盖胶质母细胞瘤和低级别胶质瘤，所有图像经过手动分割和专家审核，并经过预处理。数据集还提供了总体生存数据，用于公平比较参与方法。