2016 Glasser MMP1.0 Cortical Atlases

The human connectome project's group level multimodal parcellation (MMP; Glasser et al. [2016] Nature) is here projected from surface coordinates into volumetric space using registration fusion. 329 participants from two previously published studies (N=88 and N=241) contributed their individual alignments for registration fusion, all of which were computed by running fmriprep with default parameters and --recon-all to produce freesurfer surface segmentations and alignments. By passing the surface atlas through the subject specific transformations we obtain subject specific projections of the atlas onto two standard templates in widespread use, MNI152NLin6Asym (FSL/HCP "standard" space) and MNI152NLin2009cAsym (fmriprep's default space). Probabilistic parcel labels were derived based on individual subject alignments. Data is available as probability maps for each parcel, and also as labeled parcellations obtained using winner-take-all label assignments and a 0.2 probability threshold, in addition to individual subject segmentations in template space.File dictionary:<br>glasser_MNI152NLin[2009c|6]Asym_atlas.nii.gz - probabilistic maps projected to two templates' spaces<br>glasser_atlas_labels.txt - labels of parcels above, line order corresponds to volume order.<br>glasser_MNI152NLin[2009cA|6]Asym_labeled_p20.nii.gz - winner-takes-all parcellation derived from probabilistic maps (p &gt; 0.2) with a connectome workbench compatible color map.<br>[lr]h_[bmrk5|paingen]_MNI152NLin[2009c|6]Asym.nii.gz - subject specific parcellations projected in templates' space. One per hemisphere<br>[rl]h.HCP.MMP1 - glasser atlas resampled to fsaverage surfaces<br>methods - informal methods description (but still pretty formal)Refer to methods.txt for scanning protocols and participant details.For a comparison with https://figshare.com/articles/dataset/HCP-MMP1_0_projected_on_MNI2009a_GM_volumetric_in_NIfTI_format/3501911 please see README here:<br>https://github.com/canlab/Neuroimaging_Pattern_Masks/tree/master/Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation

群体水平多模态分区（MMP; Glasser等[2016]《自然》（Nature））源自人类连接组计划（Human Connectome Project），本数据集通过配准融合（registration fusion）技术，将该图谱从表面坐标投影至体素空间。本研究纳入两项已发表研究的329名参与者（N=88与N=241），他们提供了用于配准融合的个体配准结果。所有配准均通过默认参数运行fMRIPrep，并添加--recon-all指令以生成FreeSurfer表面分割与配准结果。通过将表面图谱经由个体专属变换传递，我们得到该图谱在两种广泛使用的标准模板空间中的个体专属投影：MNI152NLin6Asym（FSL/HCP「标准」空间）与MNI152NLin2009cAsym（fMRIPrep默认空间）。概率分区标签基于个体配准结果推导得到。数据以两种形式提供：一是各分区的概率图；二是通过赢家通吃（winner-take-all）标签分配策略结合0.2概率阈值得到的带标签分区结果，此外还包含模板空间中的个体专属分割结果。

文件说明：
- glasser_MNI152NLin[2009c|6]Asym_atlas.nii.gz：投影至两种模板空间的概率图谱
- glasser_atlas_labels.txt：上述分区的标签文件，行顺序与体素顺序一致
- glasser_MNI152NLin[2009cA|6]Asym_labeled_p20.nii.gz：由概率图（p>0.2）通过赢家通吃策略得到的带标签分区，配有兼容连接组工作台（Connectome Workbench）的颜色映射表
- [lr]h_[bmrk5|paingen]_MNI152NLin[2009c|6]Asym.nii.gz：投影至模板空间的个体专属分区，每个文件对应一个大脑半球
- [lr]h.HCP.MMP1：重采样至fsaverage表面的Glasser图谱
- methods：非正式但仍较为规范的方法说明

扫描方案与参与者详情请参考methods.txt文件。如需与数据集https://figshare.com/articles/dataset/HCP-MMP1_0_projected_on_MNI2009a_GM_volumetric_in_NIfTI_format/3501911进行对比，请参阅此处的README文件：https://github.com/canlab/Neuroimaging_Pattern_Masks/tree/master/Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation

搜集汇总

背景与挑战

背景概述

该数据集是基于2016年Glasser等人提出的多模态脑皮层分区（MMP），通过注册融合方法将表面坐标投影到体积空间，适用于神经影像分析。它包含概率图谱和标记分区，支持两个常用标准模板（MNI152NLin6Asym和MNI152NLin2009cAsym），并基于329名参与者的个体对齐数据生成，可用于脑图谱研究和认知神经科学领域。