FOMO60K
收藏资源简介:
FOMO60K是一个大规模的异构脑磁共振成像数据集,包含来自13,900个会话和11,187个受试者的60,529个脑MRI扫描。数据集由16个公开可用的数据源汇总而成,包括临床和研究级图像、多种MRI序列,以及广泛的解剖和病理变异性。为了保留原始图像特征并降低新用户的入门门槛,仅进行了最小预处理。同时提供了用于自监督预训练和微调的伴随代码。FOMO60K旨在支持大规模医疗影像中自监督学习方法的发展和基准测试。
FOMO60K is a large-scale heterogeneous brain magnetic resonance imaging (MRI) dataset. It comprises 60,529 brain MRI scans from 13,900 imaging sessions and 11,187 distinct subjects. The dataset is compiled from 16 publicly available data sources, including clinical and research-grade images, diverse MRI sequences, and a wide range of anatomical and pathological variabilities. To preserve original image characteristics and lower the entry barrier for new users, only minimal preprocessing has been applied. Accompanying code for self-supervised pre-training and fine-tuning is also provided. FOMO60K aims to support the development and benchmarking of self-supervised learning methods in large-scale medical imaging.
FOMO-60K: Brain MRI Dataset 概述
基本信息
- 许可证: fomodataset (具体条款见Data Usage Agreement)
- 标签: brain, mri, ssl, foundation_model, 3d, image
- 数据集名称: fomo-60k
- 规模: 10K<n<100K
- 任务类别: image-feature-extraction, zero-shot-classification
数据集描述
FOMO-60K 是一个大规模脑部 MRI 扫描数据集,包含临床和研究级扫描。数据集涵盖多种序列,包括:
- T1, MPRAGE, T2, T2*, FLAIR, SWI, T1c, PD, DWI, ADC 等。
数据集统计
- 受试者数量: 11,187
- 会话数量: 13,900
- 扫描数量: 60,529
数据格式
所有数据以 NIfTI 文件格式提供,经过标准化和预处理(包括去颅骨、RAS 重定向、共配准)。数据结构如下:
-- fomo-60k |-- sub_01 |-- ses_01 |-- t1.nii.gz
- 同一序列多次扫描命名为
sequence_x.nii.gz。 - 序列信息不可用时命名为
scan_x.nii.gz。
数据来源
数据集收集自以下公开来源:OASIS, BraTS, MSD, IXI, MGH Wild, NKI, SOOP, NIMH, DLBS, IDEAS, ARC, MBSR, UCLA, QTAB, AOMIC ID1000。
使用条款
使用本数据集需遵守 Data Usage Agreement 和 Citation Policy。




