[论文精读]The minimal preprocessing pipelines for the Human Connectome Project

论文原文：The minimal preprocessing pipelines for the Human Connectome Project - ScienceDirect

未完待续

1. 论文逐段精读

1.1. Abstract

        ①The Human Connectome Project (HCP) includes multiple magnetic resonance imaging (MRI) data

        ②HCP needs more preprocessing pipelines such as spatial artifact/distortion removal, surface generation, cross-modal registration and alignment to standard space

        ③CIFTI file format and the associated grayordinate spatial coordinate system is their final output

1.2. Introduction and rationale

        ①Imaging modalities contain:

a) T1-weighted (T1w) and T2-weighted (T2w) structural scans

b) resting-state and task-based functional MRI scans

c) diffusion-weighted MRI scans

        ②Even though unprocessed data can be directdly used in analysis, processed might bring better results and reduce error

        ③It is necessary for HCP in different stages to use different processing approaches

        ④The authors propose 6 minimal (i.e. minimize the amount of deleted information) pre-processing pipelines for 

a) removing spatial artifacts and distortions

b) generating cortical surfaces, segmentations, and myelin maps

c) making the data easily viewable in the Connectome Workbench visualization software

d) generating precise within-subject cross-modal registrations

e) handling surface and volume cross-subject registrations to standard volume and surface spaces

f) making the data available in the CIFTI format in a standard “grayordinate” space

consortium  n.(合作进行某项工程的)财团，银团，联营企业；联盟 (复数：consortia)

esoteric  adj.秘传的；机密的, 隐秘的；只有内行才懂的；具神秘教义的

oblique  adj.斜的；间接的，拐弯抹角的  n.倾斜物；斜线符  vi.倾斜  adv.成45度角地

1.3. The CIFTI file format and grayordinates, a combined cortical surface and subcortical volume coordinate system

        ①They think 2D geodesic distances are better than 3D Euclidean distances in that brains are always full of sulcus

        ②The 3D subcortical structure is best analyzed in 3D volume space

        ③The Connectivity Informatics Technology Initiative (CIFTI) file format stores gray matter data

        ④Standard grayordinate space is the combination of a set of standard subcortical volume parcels and the right and left standard cortical surface meshes. The picture below is CIFTI:

        ⑤Most CIFTI file includes 2D matrix, the one dimension is spatial domain (often the standard set of grayordinates). For instance, row represents spatial dimensionality and column represents time potint in dense timeseries file. However, in dense Connectome file, the two dimensionalities are spatial domain.

        ⑥Dense timeseries file:

        ⑦Dense Connectome file:

        ⑧If only need to perform independent analysis of each grayscale without referring to the spatial relationship between grayscales,  CIFTI file can be converted to  NIFTI-1 file

        ⑨NIFTI can be split into GIFTI surface data and NIFTI volume data

geodesic  adj.测地线的，大地线的(曲面上两点间距离最短的线)  n.【数】同“geodesic line”

cerebral  adj.脑的;大脑的;智力的;理智的    sulcal  adj.沟的

thalamus  n.丘脑;视丘;背侧丘脑

1.4. HCP structural acquisitions

        ①HCP includes T1-weighted (T1w) and T2-weighted (T2w) images (0.7 mm isotropic)

（在医学影像学中，T1加权和T2加权是描述MRI（核磁共振）图像中的两种不同权重类型。

T1加权（T1WI）主要反映组织纵向弛豫过程，在T1加权像上，正常组织的信号强度通常与周围液体的信号强度相近。T1加权像通常用于显示组织结构的细节和轮廓，如灰质和白质的区分。

T2加权（T2WI）则反映组织纵向弛豫的异步性，可以理解为突出组织横向弛豫过程。在T2加权像上，长T2信号通常显示为亮的改变，而短T2信号则显示为暗的改变。例如，脑脊液在T2加权像上通常显示为黑色（低信号），而灰质和白质则显示为白色（高信号）。

总的来说，T1加权和T2加权是MRI图像的两种不同的权重类型，它们分别强调了组织的纵向弛豫和横向弛豫过程，并可用于区分不同的组织结构。）

1.5. HCP functional and diffusion acquisitions

1.6. HCP pipelines: minimum acquisition requirements

1.7. HCP pipelines

1.8. HCP pipelines: overview

1.9. HCP pipelines: PreFreeSurfer

1.10. HCP pipelines: FreeSurfer

1.11. HCP pipelines: PostFreeSurfer

1.12. HCP pipelines: fMRIVolume

1.13. HCP pipelines: fMRISurface

1.14. HCP pipelines: Diffusion Preprocessing

1.15. Future directions for the HCP minimal preprocessing pipelines

2. 知识补充

3. Reference List

Glasser, M. F. et al. (2013) 'The minimal preprocessing pipelines for the Human Connectome Project', NeuroImage, vol. 80, pp. 105-124. doi: Redirecting