A Bayesian Approach to MR Tractography in the Developing Brain

描述（由申请人提供）：虽然早产儿的存活率在过去十年中稳步提高，但不良神经发育结局的发生率基本保持不变。 在美国每年出生的50万极低出生体重婴儿中，约有50%将面临运动、认知和/或行为方面的挑战。 与早产相关的主要神经病理发生在大脑白色物质（WM）中，其次影响发育中的大脑皮质。 该队列的死亡率非常低，限制了可用于研究的病理材料的量。 因此，早产儿脑WM的定量评价方法是迫切需要的。 这些方法可用于定义正常的WM发育，这将允许监测旨在优化大脑发育的新生儿干预措施，以及识别有后期认知障碍风险的婴儿。 磁共振扩散测量可以提供有关WM显微结构和神经纤维束的信息。 目前，尚不清楚哪些参数是白色物质完整性或质量的最佳指标。 同样，对于识别或遵循WM片段的最佳方法也没有达成共识。 目前，扩散张量模型是最常用的，并且通常单独应用于个体体素。 大脑中的白色物质纤维束在许多体素上延伸，并且可以通过扩展扩散张量模型以包括与相邻体素的局部连接来更好地建模。 贝叶斯概率理论为我们提供了最佳模型选择和参数估计的工具，可以更好地评估WM连接性，并为神经纤维束及其评估提供一致的概率理论基础。 候选人的长期目标是开发扩散MR成像方法，以使用贝叶斯概率理论提供WM发展和成熟的准确评估。 中心假设是贝叶斯概率理论将提供一种最佳参数估计的方法，该方法将提供关于WM连接状态的准确信息。 本申请的目的是开发基于贝叶斯分析所需的软件工具，并将其最初应用于模拟数据，然后应用于正常的离体狒狒大脑，然后研究正常的人类婴儿。 本研究将以评价离体狒狒脑中的WM损伤与组织学相关性作为结论。 贝叶斯概率理论尚未应用于WM发育的评价，候选人能够将人类新生儿结果与正常和异常的离体狒狒大脑进行比较，这是一种成熟的人脑模型。

DESCRIPTION (provided by applicant): While survival rates for premature infants have improved steadily over the last decade, the incidence of adverse neurodevelopmental outcomes has remained essentially unchanged. Approximately 50% of the half million very low birth-weight infants born each year in the United States will face motor, cognitive, and/or behavioral challenges. The principal neuropathology associated with prematurity occurs in the cerebral white matter (WM), with secondary impact on the developing cerebral cortex. The mortality rate in this cohort is very low, limiting the amount of pathological material available for study. Thus, methods for quantitative evaluation of cerebral WM in preterm infants are urgently required. Such methods could be used to define normal WM development, which would allow the monitoring of neonatal interventions aimed at optimizing cerebral development as well as identifying infants at risk for later cognitive impairment. MR diffusion measurements can provide information on WM microstructure and on neuronal fiber tracts. At present, it is not clear which parameters are the best indicators of white matter integrity or quality. Similarly, there is no consensus on the best means by which to identify or follow WM tracts. Currently, the diffusion tensor model is the most commonly used and is usually separately applied to individual voxels. White matter fiber bundles in the brain extend over many voxels and could be better modeled with extension of the diffusion tensor model to include local connectivity with neighboring voxels. Bayesian probability theory provides us with the tools for optimal model selection and parameter estimation that can better evaluate WM connectivity and provide a consistent probability theory basis for neuronal fiber tracts and their evaluation. The candidate's long-term goal is to develop diffusion MR imaging methods to provide an accurate evaluation of WM development and maturation using Bayesian probability theory. The central hypothesis is that Bayesian probability theory will provide a means for optimal parameter estimation that will provide accurate information on the status of WM connectivity. The objective in this application is to develop the software tools needed for Bayesian based analysis and to apply it initially to simulated data, followed by application to normal ex vivo baboon brain, followed by a study of normal human infants. The study will conclude with an evaluation of WM injury in ex vivo baboon brains with histological correlates. Bayesian probability theory has not been applied to the evaluation of WM development, and the candidate is able to compare human neonate results with normal and abnormal ex vivo baboon brains, which is a well-established model for human brain maturation.