Statistical Methods for Whole-Brain Dynamic Connectivity Analysis

我的K25奖项的目标是将我自己确立为一名独立的神经影像统计学家， 全脑网络分析方面的专业知识，以及致力于多学科研究团队的重要成员 解决大脑疾病的问题。要实现这些目标，需要进行教学培训和研究指导。 研究 我们将开发新的方法来改进正常和正常的全脑动态连接性分析 大脑功能异常，这对于了解各种大脑疾病，如阿尔茨海默氏症， 并可能有助于识别生物标志物，并为早期预防和治疗提供信息。以前的研究主要是 基于使用来自整个大脑扫描的数据构建的一个平均网络(即，静态连接)，但是 新出现的证据表明，网络拓扑在秒到分钟的范围内显示出有意义的变化， 除非这些差异被量化，否则就会造成理解上的差距。虽然已经有几种方法 为解决该领域的这一新方向，目前还没有一个统一的框架 准确估计全脑网络，以及跨职能部门的网络变化动态 磁共振成像(FMRI)实验，同时a)考虑感兴趣的变量和运动- 诱导伪影和b)允许动态的单独估计。这里提出的新方法将 解决这些需求，为未来的动态脑网络分析研究提供一套工具。这 研究，以及我提出的培训计划，将有助于我成为一名独立的 在大脑网络分析方面有专长的神经影像统计学家。 培训 拟议的培训计划包括四个部分：1)职业指导以及神经科学和网络 来自指导委员会的分析培训；2)建立基础的教育组成部分 具备计算神经科学和图像分析方面的知识；3)使用 从拟议的培训计划中获得的技能；4)参与知识和思想的交流 其他统计学家和神经学家通过研讨会、会议、系列研讨会和期刊俱乐部。这个 培训将使我从一名早期的职业统计员转变为一名成熟的、独立的神经成像 在全脑网络分析方面有专长的统计学家。计算神经科学方面的培训和 图像分析将使我成为一名多学科的研究团队科学家，致力于研究 人脑。在这五年中获得的增长将带来一套技能和信心， 让我更加精通我正在分析的数据背后的神经科学和生物学。这最终将是 与神经学家和临床医生进行更有效的沟通，改进研究设计，更多 知情的统计分析，并在我未来的工作中对结果进行更全面的解释。

My objective for the K25 award is to establish myself as an independent neuroimaging statistician, with expertise in whole-brain network analyses and an integral member of multidisciplinary research teams devoted to addressing diseases of the brain. Attaining these goals will require didactic training and research guidance. Research We will develop new methodology to improve whole-brain dynamic connectivity analyses of normal and abnormal brain function, which is vital for understanding various brain disorders, such as Alzheimer’s Disease, and may help identify biomarkers and inform early prevention and treatment. Previous studies are largely based on one average network constructed using data from an entire brain scan (i.e., static connectivity), but emerging evidence suggests network topology exhibits meaningful variations on the second to minute scale, creating a gap in understanding unless these variations are quantified. While several methods have been proposed to address this new direction in the field, there does not yet exist a unifying framework that accurately estimates whole-brain networks, as well as the dynamics of network change across a functional magnetic resonance imaging (fMRI) experiment, while a) accounting for variables of interest and motion- induced artifacts and b) allowing for individual estimates of dynamics. The novel methods proposed here will address these needs and provide a set of tools for future dynamic brain network analysis research. This research, along with my proposed training plan, will facilitate my progression toward becoming an independent neuroimaging statistician with expertise in brain network analysis. Training The proposed training program involves four components: 1) career guidance and neuroscience and network analysis training from a mentoring committee; 2) an educational component to establish fundamental knowledge in computational neuroscience and image analysis; 3) performing innovative research using the skills gained from the proposed training plan and; 4) participating in the exchange of knowledge and ideas with other statisticians and neuroscientists through workshops, conferences, seminar series, and journal clubs. The training will enable me to shift from an early career statistician to an established, independent, neuroimaging statistician with expertise in whole-brain network analyses. The training in computational neuroscience and image analysis will allow me to become a multidisciplinary research team scientist dedicated to studying the human brain. The growth gained through this 5-year period will lead to a skill set, and a confidence, that allows me to be more well-versed in the neuroscience and biology behind the data I am analyzing. This will ultimately lead to more effective communication with neuroscientists and clinicians, improved study design, more informed statistical analyses, and a more comprehensive interpretation of the results in my future work.