sherlock
收藏资源简介:
Sherlock 是一个自然主义功能磁共振成像(fMRI)数据集,专为特征提取等神经影像分析任务设计。该数据集旨在研究个体在观看复杂叙事刺激时及后续回忆过程中的神经活动共享结构。数据来源于一项实验,其中 16 名受试者在扫描仪内观看了总计约 50 分钟的《Sherlock》影视片段(分为两个部分:Part1 和 Part2,扫描重复时间 TR=1.5 秒),随后在扫描仪内对叙事内容进行了口头回忆。本仓库提供的是经过 fmriprep(版本 1.2.6-1)预处理后的数据镜像。数据集采用 BIDS 格式组织,主要包含以下内容:1) 预处理后的 fMRI 数据(位于 `fmriprep/sub-XX/` 目录下,包括 `anat/`、`func/` 等子目录);2) 实验事件起始时间文件(位于 `onsets/` 目录下,包含观看、回忆等不同条件的 onset 文件);3) 刺激材料(位于 `stimuli/` 目录下,包括《Sherlock》Part1 视频文件 `stimuli_Sherlock.m4v` 以及从视频中提取的文本特征文件 `video_text.npy`)。该数据集适用于研究自然情境下的脑功能、记忆编码与提取、跨被试神经表征对齐等认知神经科学问题。数据可通过配套的 `dartbrains-tools` 进行便捷加载。原始研究发表于 *Nature Neuroscience*,原始未处理数据存储于 OpenNeuro(标识符 ds002345)。数据集遵循 CC-BY-NC-4.0 许可协议。
Sherlock is a naturalistic functional magnetic resonance imaging (fMRI) dataset designed for neuroimaging analysis tasks such as feature extraction. It aims to investigate the shared neural activity structure in individuals during viewing of complex narrative stimuli and subsequent recall processes. The data comes from an experiment where 16 subjects watched approximately 50 minutes of Sherlock video clips (divided into two parts: Part1 and Part2, with a scan repetition time TR=1.5 seconds) inside a scanner, followed by verbal recall of the narrative content while still in the scanner. This repository provides a preprocessed data mirror using fmriprep (version 1.2.6-1). The dataset is organized in BIDS format and includes: 1) Preprocessed fMRI data (located in the `fmriprep/sub-XX/` directory, with subdirectories such as `anat/` and `func/`); 2) Experimental event onset files (in the `onsets/` directory, containing onset files for conditions like viewing and recall); 3) Stimulus materials (in the `stimuli/` directory, including the Sherlock Part1 video file `stimuli_Sherlock.m4v` and a text feature file `video_text.npy` extracted from the video). This dataset is suitable for studying brain function in naturalistic contexts, memory encoding and retrieval, cross-subject neural representation alignment, and other cognitive neuroscience questions. Data can be conveniently loaded using the accompanying `dartbrains-tools`. The original study was published in *Nature Neuroscience*, and the raw unprocessed data is stored on OpenNeuro (identifier ds002345). The dataset follows the CC-BY-NC-4.0 license.
数据集概述:Sherlock fMRI Dataset
- 数据集名称:Sherlock fMRI Dataset
- 许可证:CC-BY-NC-4.0(与上游OpenNeuro发布版本一致)
- 任务类别:特征提取(feature-extraction)
- 标签:fMRI, 神经影像, 自然刺激, 夏洛克, BIDS
数据集描述
该数据集是一个自然观看fMRI数据集,包含16名受试者在观看约50分钟《Sherlock》视频(分为Part1和Part2两个扫描run)后,在扫描仪内口头回忆故事内容时的脑部影像数据。采样重复时间(TR)为1.5秒。
数据来源与预处理
数据仓库镜像了最初通过DataLad分发的fmriprep预处理后数据集(原始数据地址:https://gin.g-node.org/ljchang/Sherlock),使用fmriprep版本1.2.6-1进行预处理。
数据集目录结构
fmriprep/sub-XX/ anat/ func/ figures/ log/ onsets/ Sherlock_Crop_Onsets.csv Sherlock_Recall_Scene_n50_Onsets.csv Sherlock_Watch_Scene_N50_Onsets.csv Sherlock_Segments_1000_NN_2017.xlsx stimuli/ stimuli_Sherlock.m4v # Part1视频(25分钟) video_text.npy # 文本特征
数据访问方式(通过dartbrains_tools)
python from dartbrains_tools.data import sherlock
bold = sherlock.get_file("sub-01", task="sherlockPart1", suffix="bold") confounds = sherlock.load_confounds("sub-01", task="sherlockPart1")
完整教程详见:https://dartbrains.org
引用信息
- 论文:Chen, J., Leong, Y., Honey, C. et al. Shared memories reveal shared structure in neural activity across individuals. Nat Neurosci 20, 115–125 (2017). DOI:https://doi.org/10.1038/nn.4450
- 原始原始数据:OpenNeuro数据集 ds002345(https://openneuro.org/datasets/ds002345)




