Multi-modal brain imaging analysis method for brain modeling

提出的研究计划的主要目标是发展和改进脑成像分析方法。脑成像技术的出现，如正电子发射断层扫描（PET）和功能性磁共振成像（fMRI），极大地扩大了我们对大脑功能的认识。例如，通过注射[18F]氟脱氧葡萄糖（FDG， PET最广泛使用的放射性示踪剂），我们可以定位葡萄糖快速代谢的位置，这与受试者在特定条件下扫描的神经元活动水平有关。尽管具有启发性，但典型的FDG-PET技术的局限性之一是无法模拟区域到区域的连接。换句话说，很难将大脑建模为一个网络，在这个网络中，遥远的大脑区域相互交流。*******利用FDG-PET图像的功能连通性和图论，我确定了人脑中的“信息流枢纽”。刺激（功能干扰）中枢脑区可能对大脑网络结构影响最大，并改变大脑处理信息的方式。在这里，使用fMRI将实现单独使用FDG-PET无法实现的个性化枢纽识别。在这里，我提出的研究计划的前半部分将集中在基于大型脑成像数据库的数学脑网络建模上，并通过使用非侵入性脑刺激技术的“扰动-测量”实验方法验证所构建的模型。这一建议将促进一种数学理论-实验合作的文化，类似于在物理科学中盛行的那种文化，而这种文化在脑成像领域一直缺乏。*******在基础神经科学研究中，使用PET的最大优势可能是可以监测放射性示踪剂所针对的特定神经化学物质的活动。然而，目前的方法通常需要动脉血液采样来估计有多少放射性示踪剂实际输送到大脑区域。这与动脉闭塞、出血和感染等风险相关。通过使用混合PET-MRI同时获取血流信息，我们将开发出在脑图像中准确描绘动脉的方法，从而用“基于图像的血液采样”取代真正的血液采样。我们还将开发一种方法来澄清PET结果测量变化的来源，这些变化通常被同时变化的血流所掩盖。*******总之，我的团队将开发克服PET和MR成像相关挑战的方法，包括但不限于上述困难。招收博士研究生1名，硕士研究生3名。本研究项目培养的HQPs及其研究成果将为脑成像科学的发展做出巨大贡献

The primary goal of the proposed research program is to develop and improve the brain imaging analysis methods. The emergence of brain imaging techniques such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) significantly contributed to broadening our knowledge about how the brain functions. For example, with injection of [18F]fluorodeoxyglucose (FDG; the most widely used radiotracer for PET), we can locate where glucose is rapidly metabolized which is related with the level of neuronal activity in particular conditions that the subject is scanned under. Although illuminating, one of the limitations of the typical FDG-PET technique is that one cannot model the region-to-region connectivity. In other words, it is difficult to model the brain as a network where distant brain regions are communicating with each other.*******Utilizing the functional connectivity and graph theory on FDG-PET images, I have identified “hubs of information flow” in human brain. Stimulating (functionally interfering) the hub brain region may have the most influence on the brain network structure and alter how the brain processes information. Here, use of fMRI will enable personalized hub identification that is not possible with FDG-PET alone. Here, the first half of my proposed research program will focus on modelling mathematical brain networks based on large brain imaging database and validate the constructed models by “perturbing-and-measuring” experimental approach using non-invasive brain stimulation technique. This proposal will promote a culture of mathematical theory-experiment partnership similar to that prevailing in the physical sciences which has been lacking in the field of brain imaging.*******In basic neuroscience research, perhaps the biggest advantage of using PET is that one can monitor the activities of specific neurochemicals that the radiotracer is targeting. However, current approaches often require arterial blood sampling to estimate how much radiotracer is actually delivered to the brain region. This is associated with risks including arterial occlusion, bleeding and infection. By simultaneously acquiring blood flow information using hybrid PET-MRI, we will develop methods that accurately delineate arteries in the brain images thereby replacing the real blood sampling with “image-based blood sampling.” We will also develop a method that clarifies the sources of changes in the PET outcome measures which has been often obscured by concurrently changing blood flow.*******In sum, my team will develop methods that overcome the challenges associated with PET and MR imaging, including but not limited to the above-mentioned difficulties. One PhD student and three MSc students will be recruited. HQPs trained in the proposed research program and their research outcome will greatly contribute to the advances of brain imaging science.***