功能磁共振图像(functional MRI, fMRI)的配准精度对于理解正常、患病或受伤大脑如何运作、恢复和重组具有重要的临床意义。由于脑解剖结构和功能区域间不存在精确的一一对应关系，fMRI配准精度本质上仍受到功能区域对齐程度的限制。然而，现有fMRI配准方案仅利用全脑结构信息或脑灰质功能信息，未能引入或部分引入功能信息，导致脑功能配准精度下降。本项目旨在利用全脑功能信息，开展脑白质信息辅助的功能配准方法新研究，其重点包括：(1) 提出功能连接张量算法鲁棒提取脑白质功能信号灰度和方向倾向性，以期解决脑白质功能信号易于噪声混淆且以往被弃用等问题 (2) 提出高阶动态功能连接矩阵提取脑灰质动态且具有判别性的功能特征 (3) 分开表征脑灰白质功能信号并融合配准，以期在获得全脑微分同胚形变场的前提下，消除由于脑灰白质血液动力学响应不同产生的功能伪影。

Accurate fMRI registration is critical for population studies of the operation of normal brains, the recovery of diseased brains, and the reorganization of injured brains. Due to the dissociation between functional and anatomy, the performance of fMRI registration intrinsically relies on precise alignment of functional areas. However, existing fMRI registration schemes only utilize using high-resolution structural images or gray matter (GM) functional features with no functional information or part of that, leading to the deterioration of functional areas alignment. Aiming to take advantage of whole brain functional information, this project develops a new fMRI registration study with additional information on white matter. The key points of this project are as follows: (1) White matter (WM) function signals are always susceptible to noise and abandoned in past decade, thus, we propose Functional Connection Tensors (FCTs) to robustly estimate the intensity and directional anisotropy of white matter function signals (2) We construct a high-order dynamic functional connectivity (HdFC) matrix to extract the dynamic and discriminant functional features of GM (3) We separately extract the functional features of GM and WM to eliminate the functional artifacts caused by different hemodynamic properties and obtain the whole brain diffeomorphic deformation field.