脑血容积/血管造影成像辅助的脑功能影像精确数据同化

The employment of an assumed venous blood volume fraction (BVF, i.e. V0) is universal in hemodynamic data assimilation due to ill-posedness of inverse problem. All studies so far have engaged a physiological plausible value V0 = 0.02 in region of interest (ROI) or throughout the whole brain to dispel the illconditioning problem, instead of investigating actual BVF. However, for BOLD modality, the intensity of signal is highly weighted by regional BVF. The active domains that are subject to assimilation processing are often overlapped with those areas with large blood content. In this situation, unrealistic BVF might lead to inaccurate assimilation results. In this project, we will present two empiric attempts where actual BVF is derived from other imaging modalities (MR angiography imaging and cerebral blood volume imaging) to augment the existing assimilation schemes. The influence of BVF on single-region assimilation and multi-region assimilation (namely effective connectivity) will be investigated in several experiments. Our preliminary results show that BVF has an important impact on assimilation results. As physical realistic BVF is adopted in assimilation process, more reasonable inference about hemodynamic behaviour can thus be expected. The combinative approach can provide more reliable model prediction and parameter estimation, and improve brain connectivity estimation from fMRI data that is of considerable interest in neuroimage community recently, such as Human Connectome Project (HCP) in NIH, Brain CONNECT Project in Europe, and several grants in National Basic Research Program of China (973). The efforts should be appreciated.

fMRI信号强度与成像区域内的静态血容积比率（blood volume fraction，BVF）有很强的关联性，因此也具有空间非均质性的特点。然而，先前的研究都忽略了这一点，没有对BVF进行定标，而是对于感兴趣区或全脑假定一个均一的生理学上貌似合理的BVF值，这可能引起血液动力学系统行为估计（single-region data assimilation），以及描述神经环路的动力学系统参数估计（multi-region data assimilation，动态因果模型的有效连接）的偏差。我们的研究整合脑血容积/血管造影成像数据获取的真实的静态脑血容积比率信息，发展多源数据融合的fMRI数据同化理论，真实的BVF信息的引入能够提高动力学系统的可辨识性，获得准确的脑功能和网络系统结构与行为信息。该研究能够提高现有fMRI脑功能研究的准确性，而且对于揭示大脑的生理、病理学机制有重要的意义。

fMRI 信号强度与成像区域内的静态血容积比率（blood volume fraction，BVF）有很强的关联性，因此也具有空间非均质性的特点。然而，先前的研究都忽略了这一点，没有对BVF 进行定标，而是对于感兴趣区或全脑假定一个均一的生理学上貌似合理的BVF 值，这可能引起血液动力学系统行为估计（single-region data assimilation），以及描述神经环路的动力学系统参数估计（multi-region data assimilation，动态因果模型的有效连接）的偏差。我们的研究整合脑血容积/血管造影成像数据获取的真实的静态脑血容积比率信息，发展多源数据融合的fMRI 数据同化理论，真实的BVF 信息的引入能够提高动力学系统的可辨识性，获得准确的脑功能和网络系统结构与行为信息。主要研究成果包括：发展了两种分别基于血管造影成像和血容积造影成像的脑血容积快速计算方法，可以实现对脑血容积比率信息的快速准确提取；从模型可辨识性的角度探讨了血容积比率对脑因果网络连接的动态因果模型的影响；发展了以最小描述长度原理为指导原则的脑因果网络连接一致性模型选择方法。该研究能够提高现有fMRI 脑功能研究的准确性，为因果脑网络的研究的发展提供指导，同时对于揭示大脑的生理、病理学机制有重要的意义。