A Toolkit for Analysis and Visualization of Preclinical Rodent Neuroimaging Experiments

项目摘要 啮齿动物模型在脑部疾病和紊乱的临床前研究以及基础研究中仍然是重要的 神经科学研究。通过核磁共振和显微镜等技术获得的啮齿动物成像数据， 在许多这样的研究中扮演着关键的角色。虽然在软件开发方面付出了大量的努力 分析大脑核磁共振的工具，这项工作的大部分都集中在人类数据上，研究人员 研究疾病的动物模型必须经常采用这些工具来进行研究。在这个项目中，我们 将通过开发一套专门的开源软件工具来处理、分析、 以及可视化从啮齿动物大脑获得的神经成像数据。这些工具将在结构性扩散上运行， 以及功能磁共振成像，以及光学清除的连续切片组织样本的光学显微镜。这些 工具将建立在我们几十年来开发用于分析人类和老鼠图像的软件经验的基础上 数据，我们开发小鼠大脑多模式图谱的经验，以及我们在社区中的积极努力 在神经科学研究中应用这些资源时的参与和传播。在适当的情况下，我们 将利用深度学习方法生成功能强大的分割和注册网络 手动批注和描绘的数据。我们还将开发易于使用的界面，以方便数据 处理并提供数据集的高级可视化功能，数据集的大小约为一万亿像素。 该项目有五个具体目标。AIM 1将开发MRI处理工具，其中将包括子对象内联合 MRI模式的注册，从整个头部扫描中提取脑组织，组织分类，以及 磁共振弥散和功能成像数据的处理。AIM 2将开发用于处理透明的显微镜的工具 和切片组织，主要目标是将这些数据与从任一光学元件生成的参考图谱对齐 显微镜或核磁共振。这些工具将在自动分割的区域进行细胞计数；轴突跟随； 树突枝状；树突棘计数。在目标3中，我们将开发一个统计分析工具箱，它 将对来自显微镜和MRI数据的神经成像测量执行统计推断，使用 在目标4中，我们将把来自目标1-3的组件集成到一个信息学平台中 它将提供用于轻松编写脚本的命令行工具、与相关映像工具的互操作性以及图形化的 用于跨不同比例可视化数据的界面。在目标5中，我们将对我们的软件工具进行评估 使用两项研究：对实验性自身免疫性脑脊髓炎(EAE)小鼠模型进行成像 硬化症；以及想象正常衰老的小鼠模型。我们还将与一个小动物网络合作 项目外部的成像专家，他们将使用和评估该软件。这些带动项目将服务于 作为试验床，确保软件在科研环境中的实用性，为 发展我们的研究平台。我们将在开放源码许可下自由分发软件，并且 通过我们的网站提供用户支持。

Project Summary Rodent models remain an important in preclinical studies of brain disease and disorders, as well as basic neuroscience investigations. Rodent imaging data, acquired through techniques including MRI and microscopy, play a critical role in many of these studies. While a great deal of effort has gone into the development of software tools for analyzing MRI of the brain, much of this work has been focused largely on human data, and investigators studying animal models of disease must often resort to adapting these tools for their research. In this project, we will address this need by developing a dedicated suite of open source software tools for processing, analyzing, and visualizing neuroimaging data acquired from rodent brains. These tools will operate on structural diffusion, and functional MRI, as well as optical microscopy of optically cleared serially sectioned tissue samples. These tools will build upon our decades of experience developing software for analyzing human and mouse imaging data, our experience in developing multimodal atlases of the mouse brain, and our active efforts in community engagement and dissemination while applying these resources in neuroscientific studies. Where suitable, we will make use of deep learning methods to produce powerful segmentation and registration networks trained on manually annotated and delineated data. We will also develop easy-to-use interfaces that will facilitate data processing and provide advanced visualization capabilities of datasets with sizes on the order of one terapixel. The project has five specific aims. Aim 1 will develop MRI processing tools, which will include intrasubject co- registration of MRI modalities, extraction of brain tissue from whole head scans, tissue classification, and processing of diffusion and functional MRI data. Aim 2 will develop tools for processing microscopy of cleared and sectioned tissue, with the major goal of aligning these data to a reference atlas generated from either optical microscopy or MRI. These tools will perform cell counting in automatically segmented regions; axon following; dendritic arborization; and dendritic spine counting. In Aim 3, we will develop a statistical analysis toolbox, which will perform statistical inference for neuroimaging measures from microscopy and MRI data analyzed using methods from Aims 1 and 2. In Aim 4, we will integrate the components from Aims 1-3 into an informatics platform that will provide command line tools for easy scripting, interoperability with related imaging tools, and a graphical interface for visualizing data across different scales. In Aim 5, we will perform evaluation of our software tools using two studies: imaging an experimental autoimmune encephalomyelitis (EAE) mouse model of multiple sclerosis; and imaging a mouse model of normative aging. We will also work with a network of small animal imaging experts external to the project, who will use and evaluate the software. These driving projects will serve as testbeds to ensure the practical utility of the software in a research setting, providing direction for the development of our research platform. We will distribute the software freely under an open source license, and provide user support through our website.