Arterial input function Independent Measures of Perfusion with Physics Driven Models

摘要 在美国，急性缺血性中风(AIS)每年影响大约70万名患者。[本杰明 尽管静脉溶栓剂的引入改善了患者的预后， 有效的机械血栓清除治疗方案的发展已经显著改变了 AIS患者的临床管理，特别是在选择合适的患者进行干预时。这个 目前的治疗选择方法利用的是患者特有的数据，严重依赖于定量神经成像 方法，要么来自计算机断层扫描(CT)，要么来自较小程度的磁共振 成像(MRI)。CT在美国相对容易获得，已成为治疗中风的主要手段 病人分诊。 脑血流灌注成像是评估缺血半暗带和梗死灶的核心。 动脉内溶栓患者选择的精确度。典型的动态CT灌注扫描 到达时，使用碘化造影剂进行40到60个时间点的重复扫描 在急诊室。这些图像被自动或半自动地后处理成灌注 指标，使用FDA批准的多个软件包。这些包基本上都依赖于类似的 动态采集图像的后处理路径，包括运动校正、动脉输入 函数选择和某种形式的反卷积后处理。生成一组灌注图， 通常包括脑血流量(CBF)、脑血容量(CBV)、平均通过时间(MTT)和时间 至最大对比度浓度(Tmax)。然后，软件包将阈值应用于CBF和 Tmax映射从CBF生成假定的缺血“核心”，并从Tmax生成“半暗带”。然而， 这些值对所选择的动脉输入函数(AIF)的依赖导致了广泛的努力 自动选择AIF，或探索系统的方法来产生局部AIF以改善血流灌注 测量。定义独立于AIF选择的灌注度量可以显著改善 卒中血流灌注分析，减少患者的辐射暴露。这项研究的目标是评估一个物理学 基于CT灌注模式的脑灌注模型评价脑灌注参数这个 新技术的关键要求包括独立于AIF选择、定量和稳定 在临床上与卒中结局相关并可预测其预后的血流灌注测量。

ABSTRACT Acute Ischemic Stroke (AIS) affects approximately 700,000 patients each year in the United States. [Benjamin EJ 2019, Circulation] Though the introduction of intravenous thrombolytics improved patient outcomes, the development of effective treatment regimens with mechanical thrombectomy has significantly altered the clinical management of AIS patients, especially when appropriate patients are selected for intervention. The current treatment selection approaches utilize patient specific data heavily relies on quantitative neuroimaging approaches, derived from either Computer Tomography (CT), or to a lesser extent magnetic resonance imaging (MRI). CT, with its relative availability within the US, has been the primary modality used for stroke patient triage. Brain perfusion imaging has been central to the evaluation of the ischemic penumbra and infarct core enabling precision in patient selection for intra-arterial thrombolysis. Typically dynamic CT perfusion scans with repeated scans 40 to 60 time points with the administration of iodinated contrast are obtained upon the arrival in the emergency room. These images are automatically or semi-automatically post-processed into perfusion metrics, using a number of FDA approved software packages. These packages all essentially rely on a similar post-processing pathway for the dynamically acquired images, consisting of motion correction, arterial input function selection and some form of deconvolution post-processing. A set of perfusion maps are generated, typically including cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT) and time to the maximum contrast concentration (Tmax). The software packages then apply thresholds to the CBF and Tmax maps to generate a presumed ischemic “core” from the CBF and “penumbra” from the Tmax. However, the dependence of these values on the arterial input function (AIF) selected has resulted in extensive efforts to automate AIF selection, or explore systematic methods to produce local AIFs to improve perfusion measurements. Defining a perfusion metric that is independent of AIF selection could substantially improve stroke perfusion analysis, and reduce patient radiation exposure. The goal of this study is to evaluate a physics based model of cerebral perfusion for evaluating perfusion parameters from CT perfusion modalities. The critical requirements of the new technique include independence from AIF selection, quantitative and stable measurements of perfusion that are clinically relevant and predictive of stroke outcomes.