Bridging diffusion MRI and chemical tracing for validation and inference of fiber architectures

连接扩散 MRI 和化学示踪以验证和推断纤维结构

• 批准号: 10318985
• 负责人: Anastasia Yendiki
• 金额: $ 65.99万
• 依托单位: MASSACHUSETTS GENERAL HOSPITAL
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-12-15 至 2024-11-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10318985
• 关键词: Address; Algorithmic Analysis; Algorithms; Anatomy; Animals; Architecture; Area; Axon; Brain; Brain Diseases; Brain imaging; Brain region; Brain scan; Caliber; Chemicals; Collaborations; Communities; Data; Data Analyses; Data Set; Databases; Development; Diffusion; Diffusion Magnetic Resonance Imaging; Disease; Engineering; Error Sources; Exhibits; Fiber; Goals; Histologic; Histology; Human; Imaging Techniques; Injections; Knowledge; Label; Link; MRI Scans; Macaca; Magnetic Resonance Imaging; Methodology; Methods; Modeling; Monkeys; Myelin; Outcome; Pathway interactions; Pattern; Performance; Prefrontal Cortex; Process; Property; Protocols documentation; Research; Research Personnel; Resolution; Resources; Role; Sampling; Seeds; Signal Transduction; Site; Source; Techniques; Therapeutic Intervention; Tracer; Training; Universities; Validation; Work; anatomical tracing; automated algorithm; brain pathway; cingulate cortex; convolutional neural network; data acquisition; deep learning; design; frontal lobe; histological stains; improved; in vivo; insight; multimodality; neuroimaging; next generation; nonhuman primate; novel; novel strategies; reconstruction; sample fixation; synergism; targeted treatment; tool; tractography

Project summary: This project will collect a unique, multi-modal, multi-contrast dataset with tracer injections and diffusion MRI in the same macaque brains. We will use this dataset to develop novel algorithms for inferring local fiber architectures from diffusion MRI. The goal is to overcome the limitations of current methods for diffusion orientation reconstruction, which are designed to resolve fiber crossings but not to distinguish between crossings and other configurations, such as branching, turning, fanning, etc. More broadly, the proposed dataset will allow us to investigate organizational principles of brain pathways and to provide a testbed for the neuroimaging community to evaluate the accuracy of diffusion tractography and microstructural modeling techniques. The project is a collaboration between groups with extensive expertise in diffusion MRI methodological development (MGH Martinos Center) and anatomical tracer studies (University of Rochester). We have previously collected high-resolution ex vivo diffusion MRI data on a set of macaque brains that had also received tracer injections. We have recently used these data in an open tractography challenge, with the participation of research teams from around the world. This was the first challenge of its kind to provide diffusion MRI data suitable for all state-of-the-art diffusion reconstruction methods (e.g., multi-shell or Cartesian grid sampling), in addition to providing the tracer injections in the same brains as the MRI scans. Our own preliminary studies and the challenge itself offer several insights into the performance of state-of-the-art tractography methods. For example, our results indicate that, while most tractography methods would require their parameters to be tuned differently to achieve optimal accuracy for different cortical seed regions, there are approaches that are robust across cortical areas. Furthermore, our results suggest that errors occur frequently in areas where the fiber architecture is not well modeled by a crossing. Thus there is a need for novel tractography approaches that go beyond the crossing-fiber paradigm. Here we propose to develop such an approach. Our prior work included injection sites in the frontal, prefrontal, and cingulate cortices only. Here we propose to investigate the extent to which our prior findings generalize across the brain, by performing tracer injections that sample a wider range of cortical areas. Furthermore, we will extend our acquisition protocol to acquire data appropriate not only for tractography, but also for microstructural and myelin mapping. These data will allow us to answer a broader range of questions about tractography, microstructure, and their intersection. Beyond the methodological development proposed in this project, the data will also be an invaluable resource to the neuroimaging community, providing researchers with a framework for the objective assessment of current diffusion MRI analysis methods and identifying areas for improvement to guide the development of next-generation techniques.

项目摘要：该项目将通过示踪剂注射收集独特的、多模式、多对比的数据集 以及在同一只猕猴大脑中进行的扩散磁共振成像。我们将使用该数据集来开发新颖的算法 从扩散 MRI 推断局部纤维结构。目标是克服当前方法的局限性 用于扩散方向重建，旨在解决纤维交叉但不区分 交叉点和其他配置之间，例如分支、转弯、扇形等。更广泛地说， 拟议的数据集将使我们能够研究大脑通路的组织原理并提供 神经影像学界评估扩散纤维束成像和微观结构准确性的测试平台 建模技术。该项目是在扩散 MRI 领域拥有丰富专业知识的团队之间的合作 方法开发（MGH Martinos 中心）和解剖示踪研究（罗切斯特大学）。 我们之前收集了一组猕猴大脑的高分辨率离体扩散 MRI 数据，这些数据 还接受了示踪剂注射。我们最近在开放式纤维束成像挑战中使用了这些数据， 来自世界各地的研究团队的参与。这是此类挑战中的第一个 扩散 MRI 数据适用于所有最先进的扩散重建方法（例如，多壳或笛卡尔 网格采样），此外还在与 MRI 扫描相同的大脑中提供示踪剂注射。我们自己的 初步研究和挑战本身为最先进的性能提供了一些见解 纤维束成像方法。例如，我们的结果表明，虽然大多数纤维束成像方法需要 它们的参数要进行不同的调整，以达到不同皮质种子区域的最佳精度， 是跨皮质区域有效的方法。此外，我们的结果表明发生了错误 经常出现在光纤架构不能很好地通过交叉建模的区域。因此需要 超越交叉纤维范式的新颖纤维束成像方法。在这里我们建议开发这样的 一种方法。我们之前的工作仅包括额叶、前额叶和扣带皮层的注射部位。这里 我们建议通过执行以下操作来调查我们先前的发现在整个大脑中的推广程度 示踪剂注射可对更广泛的皮质区域进行采样。此外，我们将扩大收购范围 协议获取的数据不仅适用于纤维束成像，而且适用于微结构和髓磷脂绘图。 这些数据将使我们能够回答有关纤维束成像、微观结构及其特征的更广泛的问题。 路口。除了本项目提出的方法开发之外，数据也将是 神经影像学界的宝贵资源，为研究人员提供了实现目标的框架 评估当前的扩散 MRI 分析方法并确定需要改进的领域以指导 下一代技术的开发。