Combined Segmentation and Hemodynamic Analysis of Cerebrovascular Structures using Spatiotemporal Arterial Spin Labeling MRI

磁共振成像（MRI）已成为一种不可缺少的诊断和研究工具。它能够可视化器官的解剖结构和功能。由于MRI技术的进步，时间分辨率三维（即4D） MRI血流测量，例如动脉自旋标记（ASL）磁共振血管造影（MRA），最近越来越受到人们的关注。目前，4D ASL MRA序列能够成像脑血管结构和血流，空间分辨率优于1mm3，时间分辨率低于100ms。由于该成像技术利用血液作为内在造影剂，因此不需要外源性造影剂，使其便宜且安全。然而，4D ASL MRA获得的大量数据量限制了该技术的实用性，特别是如果仅有的2D图像仅用于视觉检查。本研究的广泛目标是为大时间分辨率数据集开发新的图像分析算法和可视化技术，以实现血流分析、血管分割和适合4D ASL MRA数据集的血流动力学可视化技术，从而实现快速、准确和定量的解释。******我们将在第一步设计、优化和评估各种隐式和显式模型，用于基于每个体素的ASL时间强度曲线的血流动力学分析。需要对脑血管系统进行分割，以定量地研究血管解剖并将结果可视化。因此，我们将开发新的先进的血管分割方法，从大脑的4D ASL MRA数据集中提取血管。更具体地说，我们将血流动力学分析和脑血管分割结合成一个综合分析方法。期望这种血管分割和血流分析的耦合将使分割和血流动力学分析结果都得到改进。最后，将开发和评估用于脑血管系统及其血流组合表示的先进可视化方法。例如，计划设计基于字形的四维血流可视化以及基于动态表面的可视化技术。******所有方法将集成在一个新颖的软件工具中，该软件工具将提供给应用成像研究人员，从而提高4D ASL MRA的实用价值。总的来说，这项工作将能够快速、准确和定量地解释4D ASL MRA数据集，同时为其他图像处理问题创建新的图像分析和可视化方法，并具有广泛的应用。

Magnetic resonance imaging (MRI) has become an indispensable diagnostic and research tool. It is capable of visualizing the anatomy and function of organs. Time-resolved three-dimensional (i.e. 4D) MRI blood flow measurements, e.g. arterial spin labeling (ASL) MR angiography (MRA), of the brain is of increasing recent interest due to MRI technology advancements. 4D ASL MRA sequences are now capable of imaging cerebrovascular structures and blood flow with a spatial resolution better than 1 mm3 and a temporal resolution below 100 ms. Since this imaging technique utilizes blood as an intrinsic contrast agent, no exogenous contrast media is required, making it inexpensive and safe. However, the vast data volume obtained with 4D ASL MRA limits the utility of the technique, particularly if the only 2D images are only visually inspected. The broad goal of this research is to develop novel image analysis algorithms and visualization techniques for large time-resolved datasets to achieve blood flow analysis, vessel segmentation, and hemodynamic visualization techniques tailored to 4D ASL MRA datasets to enable fast, accurate, and quantitative interpretation.******We will design, optimize, and evaluate various implicit and explicit models for the hemodynamic analysis of blood flow dynamics based on the ASL time-intensity curves for each voxel in a first step. A segmentation of the cerebrovascular system is required to investigate the vessel anatomy quantitatively and to visualize the results. Therefore, we will develop new advanced vessel segmentation methods for extraction of the vessels from 4D ASL MRA datasets of the brain. More specifically, we will combine the hemodynamic analysis and cerebrovascular segmentation into an integrated analysis approach. It is expected that such a coupled vessel segmentation and blood flow analysis will lead to both improved segmentation and hemodynamic analysis results. Finally, advanced visualization methods for the combined representation of the cerebrovascular system and its blood flow will be developed and evaluated. For example, it is planned to design a glyph-based visualization of the 4D blood flow as well as dynamic surface-based visualization techniques. ******All methods will be integrated within a novel software tool that will be made available for applied imaging researchers, thereby enhancing the practical value of 4D ASL MRA. Collectively, this work will enable a fast, accurate, and quantitative interpretation of 4D ASL MRA datasets while creating new image analysis and visualization methods with broad applications for other image processing problems.