NOD-EEG

# Summary

The human brain can rapidly recognize meaningful objects from natural scenes encountered in everyday life. Neuroimaging with large-scale naturalistic stimuli is increasingly employed to elucidate these neural mechanisms of object recognition across these rich and daily natural scenes. However, most existing large-scale neuroimaging datasets with naturalistic stimuli primarily rely on functional magnetic resonance imaging (fMRI), which provides high spatial resolution to characterize spatial representation patterns but is limited in capturing the temporal dynamics inherent in visual cognitive processing.

To address this limitation, we extended our previously collected Natural Object Dataset-fMRI (NOD-fMRI) by collecting both magnetoencephalography (MEG) and electroencephalography (EEG) data from the same subjects while viewing the same set of naturalistic stimuli. As a result, the NOD uniquely integrates three different modalities—fMRI, MEG, and EEG—thus offering promising avenues to examine brain activity induced by naturalistic stimuli with both high spatial and high temporal resolutions. Additionally, the NOD encompasses a diverse array of naturalistic stimuli and a broader subject pool, enabling researchers to explore differences in neural activation patterns across both stimuli and subjects.

We anticipate that the NOD dataset will serve as a valuable resource for advancing our understanding of the cognitive and neural mechanisms underlying object recognition.

The MEG data's accession number is `ds005810`.

---

# Data Records

## Directory Structure

The raw data from each subject are stored in the `sub-subID` directory, while preprocessed data and epoch data are stored in the following directories:
- **Preprocessed Data:** `derivatives/preprocessed/raw`
- **Epoch Data:** `derivatives/preprocessed/epochs`

### Stimulus Images

The stimulus images used for MEG and EEG are identical and are stored in the `stimuli/ImageNet` directory. Images within this folder are named in the `synsetID_imageID.JPEG` Where:
- `synsetID` is the ILSVRC category information.
- `imageID` is the unique number for the image within that category.

The image metadata, including category information, is available in the table files under the `stimuli/metadata` directory.

### Raw Data

Raw EEG data are stored in BIDS format. Each subject's directory contains multiple session folders, designated as `ses-sesID`. Comprehensive trial information for each subject is documented in the file: `derivatives/detailed_events/sub-subID_events.csv` Where each row corresponds to a trial, and each column contains metadata for that trial, including the session and run number, category information of the stimuli, and subject response.

### Preprocessed Data

The full time series data of preprocessed data are archived in the `derivatives/raw` directory, named as: `sub-subID_ses-sesID_task-ImageNet_run-runID_eeg_clean.fif`. The epoch data derived from preprocessed data are stored within the `derivatives/epochs` directory. In this directory, all data for each subject are concatenated into a single file, labeled as: `sub-subID_epo.fif`

The trial information within each subject's epochs data can be accessed via the metadata of the epochs data, which are aligned with the content of the subject's `sub-subID_events.csv` file.

---

# References
Appelhoff, S., Sanderson, M., Brooks, T., Vliet, M., Quentin, R., Holdgraf, C., Chaumon, M., Mikulan, E., Tavabi, K., Höchenberger, R., Welke, D., Brunner, C., Rockhill, A., Larson, E., Gramfort, A., and Jas, M. (2019). MNE-BIDS: Organizing electrophysiological data into the BIDS format and facilitating their analysis. *Journal of Open Source Software, 4*(1896). [https://doi.org/10.21105/joss.01896](https://doi.org/10.21105/joss.01896)

# 摘要
人类大脑可从日常所见的自然场景中快速识别出具有语义的物体。借助大规模自然刺激的神经成像技术正被愈发广泛地应用，以阐明在这类丰富且贴近日常的自然场景中，物体识别的神经机制。然而，现有多数搭载自然刺激的大规模神经成像数据集主要依赖功能磁共振成像（functional magnetic resonance imaging, fMRI），该技术虽可提供高空间分辨率以刻画空间表征模式，但在捕捉视觉认知加工中固有的时间动态特性方面存在局限。

为弥补这一局限，我们对此前采集的自然物体数据集-fMRI（Natural Object Dataset-fMRI, NOD-fMRI）进行了拓展：在同一组被试观看相同自然刺激集的过程中，额外采集了脑磁图（magnetoencephalography, MEG）与脑电图（electroencephalography, EEG）数据。由此，本自然物体数据集（Natural Object Dataset, NOD）首次整合了fMRI、MEG与EEG三种模态数据，为探究兼具高空间与高时间分辨率的自然刺激所诱发的脑活动提供了极具潜力的研究路径。此外，NOD涵盖了多样化的自然刺激集与更广泛的被试池，使得研究者能够探索不同刺激与不同被试间的神经激活模式差异。

我们期望本数据集能够成为推动我们理解物体识别背后认知与神经机制的宝贵研究资源。

本数据集的脑磁图数据获取编号为`ds005810`。

---

# 数据记录
## 目录结构
每名被试的原始数据存储于`sub-subID`目录下，预处理数据与分段数据则存储于以下目录：
- **预处理数据**：`derivatives/preprocessed/raw`
- **分段数据**：`derivatives/preprocessed/epochs`

### 刺激图像
用于MEG与EEG实验的刺激图像完全一致，均存储于`stimuli/ImageNet`目录下。该文件夹内的图像命名格式为`synsetID_imageID.JPEG`，其中：
- `synsetID`为ILSVRC分类类别信息
- `imageID`为该类别内单张图像的唯一编号

包含分类信息在内的图像元数据，可于`stimuli/metadata`目录下的表格文件中获取。

### 原始数据
原始EEG数据以BIDS格式存储。每名被试的目录下包含多个会话文件夹，命名为`ses-sesID`。每名被试的完整试次信息记录于`derivatives/detailed_events/sub-subID_events.csv`文件中：该文件每一行对应一次试次，每一列则包含该试次的元数据，包括会话与运行编号、刺激的分类信息以及被试的反应。

### 预处理数据
预处理后的完整时间序列数据归档于`derivatives/raw`目录下，命名格式为：`sub-subID_ses-sesID_task-ImageNet_run-runID_eeg_clean.fif`。由预处理数据生成的分段数据则存储于`derivatives/epochs`目录中：该目录下每名被试的所有数据会被合并为单个文件，命名为`sub-subID_epo.fif`。

每名被试分段数据中的试次信息，可通过分段数据的元数据获取，该元数据与对应被试的`sub-subID_events.csv`文件内容完全对齐。

---

# 参考文献
Appelhoff, S., Sanderson, M., Brooks, T., Vliet, M., Quentin, R., Holdgraf, C., Chaumon, M., Mikulan, E., Tavabi, K., Höchenberger, R., Welke, D., Brunner, C., Rockhill, A., Larson, E., Gramfort, A. 与 Jas, M. (2019). MNE-BIDS：将电生理数据整理为BIDS格式并助力分析。 *Journal of Open Source Software*, 4(1896). [https://doi.org/10.21105/joss.01896](https://doi.org/10.21105/joss.01896)

搜集汇总

背景与挑战

背景概述

NOD-EEG数据集是一个多模态神经影像数据集，专注于研究自然场景中物体识别的神经机制。它包含了19名参与者的EEG数据，以及相应的刺激图像和预处理数据，为研究高时空分辨率的脑活动提供了宝贵资源。