Computational Tools for Modeling Human and Mouse Connectome with Multi-Shell Diffusion Imaging

摘要 弥散磁共振成像(Dmri)是在体脑连接体研究的重要工具。 灵敏的成像标记，并允许通过纤维束造影术检查连接路径。随着成功的到来 在人类连接组计划(HCP)中，高分辨率、多壳层扩散成像正在成为 连接组研究中dmri数据采集的标准方法。充分释放多壳潜力 在这个项目中，我们将开发一套新的计算工具来联合估计纤维的取向。 分布(FOD)和隔室参数。使用基于FOD的跟踪成像，我们可以可靠地解决 交叉纤维和重建纤维束，忠实地遵循已知的解剖学，如视网膜病变 视觉路径。房室参数为研究局部细胞提供了灵敏的成像标志 轴突周围的环境。我们的工具一般适用于人和小鼠的连接体 研究。弥散纤维束成像的一个主要挑战是缺乏严格的生物学验证。 有意义的基本事实。使用来自Mouse Connectome的小鼠大脑的大规模示踪剂注射数据 南加州大学MCP项目和艾伦老鼠脑连接图谱，我们将进行系统验证 和我们基于FOD的技术的优化，从成像信号的去噪到配置 小室模型对光固化成像参数的选择。这将创建一个经过良好验证的系统，用于 用多壳成像研究小鼠连接体，为人类的设计提供直观的指导 基于FOD的连接体的研究。我们的项目有三个具体目标：1.开发一个通用的 多壳纤维取向和隔室模型联合估计的计算框架 扩散成像。2.利用小鼠多层dMRI数据对计算工具进行验证和优化 从示踪剂注射得到的大脑和轴突投影图。3.开发光纤综合工具包 基于人脑多层dmri数据的束重建。在这个项目中，我们将应用我们的软件 分析在两项疾病研究中收集的数据的工具。在第一项研究中，我们将重点研究皮质纹状体-- 在BTBD3小鼠模型中检测丘脑-皮质(CSTC)网络的连通性变化 强迫症(OCD)。在第二个研究中，我们将运用我们的工具来研究 视网膜损伤和视觉通路的完整性，通过视网膜复制-保护视觉区域之间的连通性。 在该项目中开发的所有软件工具将免费分发给研究社区。

ABSTRACT For the in vivo investigation of brain connectome, diffusion MRI (dMRI) is an important tool as it provides highly sensitive imaging markers and allows the examination of connection paths via tractography. With the success of the Human Connectome Project (HCP), high resolution, multi-shell diffusion imaging is emerging as the standard approach for dMRI data acquisition in connectome studies. To fully unleash the potential of multi-shell dMRI, in this project we will develop a suite of novel computational tools that jointly estimate fiber orientation distributions (FOD) and compartmental parameters. With FOD-based tractography, we can reliably resolve crossing fibers and reconstruct fiber bundles that faithfully follow known anatomy such as the retinotopy of visual pathways. Compartmental parameters provide sensitive imaging markers for studying local cellular environment surrounding the axons. Our tools are generally applicable for both human and mouse connectome research. One main challenge in diffusion tractography is the lack of rigorous validations with biologically meaningful ground truth. With large-scale tracer injection data of mouse brains from the Mouse Connectome Project (MCP) at USC and the Allen Mouse Brain Connectivity Atlas, we will perform a systematic validation and optimization of our FOD-based techniques from the denoising of imaging signals to the configuration of compartment models to the selection of tractography parameters. This will create a well-validated system for studying mouse connectome with multi-shell imaging, and provide intuitive guidelines for the design of human studies with FOD-based connectome. There are three specific aims in our project: 1. To develop a general computational framework for the joint estimation of fiber orientation and compartment models from multi-shell diffusion imaging. 2. To validate and optimize the computational tools using multi-shell dMRI data of mouse brains and axonal projection maps from tracer injections. 3. To develop a comprehensive toolkit for fiber bundle reconstruction from multi-shell dMRI data of human brains. In this project we will apply our software tools to analyze data collected in two disease studies. In the first study, we will focus on the cortico-striato- thalamo-cortical (CSTC) network and examine its connectivity changes in the BTBD3 mouse model of obsessive-compulsive disorder (OCD). In the second study, we will apply our tools to study the relation of retinal impairment and visual pathway integrity via the retinotopy-preserving connectivity between visual areas. All software tools developed in this project will be distributed freely to the research community.