A New Multimodal Perfusion Imaging Method Using Concurrent NIRS and fMRI

描述（由申请人提供）：这个R21项目将开发和验证一种新的体内血流动力学参数成像方法（CBV和CBF），该方法基于同时记录的fMRI和NIRS数据的跨模态处理，使用渐进时间延迟（RIPTiDe）的回归插值。这种新的非侵入性成像技术允许与BOLD成像数据同时常规生成定量CBF和CBV图像，并避免了现有方法（DSC MRI， ASL和VASO）的许多缺陷。背景：RIPTiDe成像利用了近红外光谱和功能磁共振成像都测量血氧和浓度波动的事实，但没有仪器噪声机制。因此，NIRS和BOLD数据的时间互相关代表了血液氧合和容量内源性波动通过脉管系统传播的强度和时间。我们可以在高信噪比的短暂扫描中量化这些信号的幅度和到达时间(1)，并使用该技术从BOLD数据中过滤生理噪声(2)，并测量脑血管对屏气挑战的反应性(3)。利用现有的BOLD效应生物物理模型，我们建议使用这些数据在高空间分辨率下定量估计脑血流量和容量。测量可以与传统的功能磁共振成像采集同时进行，不需要特殊的功能磁共振成像采集序列或参数。此外，这种类型的测量所需的近红外采集硬件原则上可以相当便宜，使其添加到现有的磁共振扫描仪是可行的。我们将减少这种测量实践，并将其结果和数据质量与ASL和VASO进行比较。意义：使用RIPTiDe处理的同时获得的fMRI/NIRS数据使我们能够在每个体素中分离出血流动力学波动对BOLD信号的贡献。这可以显著减少BOLD数据中的生理噪声，同时产生每个位置的血流量和体积估计值。这允许真正并发获取高质量的BOLD、CBV和CBF信息。具体目的：1)比较使用近红外光谱从四个不同的记录位置获得的数据质量。近红外光谱记录的位置影响被测血流动力学信号的纯度。将使用来自四个探头位置的近红外光谱数据计算RIPTiDe图像，并选择一个标准记录位置进行比较；2)评估使用RIPTiDe数据作为气球模型的输入，以产生血流量和体积的定量估计。我们将使用RIPTiDe数据（最佳延迟NIRS [HbR]和[tHb]，以及BOLD）作为气球模型的输入，计算CBV和CBF，并将结果和信噪比/单位时间与ASL和VASO进行比较；3)实现RIPTiDe处理包。该目标将开发一套简化的数据记录和分析套件，以简化RIPTiDe数据的使用，并允许其他研究人员使用这种方法。

DESCRIPTION (provided by applicant): This R21 project will develop and validate a novel method of imaging hemodynamic parameters (CBV and CBF) in vivo, based on crossmodal processing of concurrently recorded fMRI and NIRS data using Regressor Interpolation at Progressive Time Delays (RIPTiDe). This new, non-invasive imaging technique allows the routine generation of quantitative CBF and CBV images simultaneously with BOLD imaging data, and avoids many pitfalls of existing methods (DSC MRI, ASL and VASO). Background: RIPTiDe imaging exploits the fact the NIRS and fMRI both measure blood oxygenation and concentration fluctuations, but share no instrumental noise mechanisms. Therefore the temporal crosscorrelation of the NIRS and BOLD data represents the strength and timing of the propagation of endogenous fluctuations in blood oxygenation and volume through the vasculature. We can quantify the amplitude and arrival time of these signals in single subjects during brief scans at high signal to noise(1), and have used this technique for filtering physiological noise from BOLD data (2), and measuring cerebrovascular reactivity to a breathhold challenge (3). Using existing biophysical models of the BOLD effect, we propose using this data to quantitatively estimate cerebral blood flow and volume at high spatial resolution. Measurements can be made concurrently with conventional fMRI acquisitions, and require no special fMRI acquisition sequences or parameters. Moreover, the near infrared acquisition hardware required for this type of measurement can in principal be quite inexpensive, making it practical to add it to existing MR scanners. We will reduce this measurement to practice, and compare its results and data quality to ASL and VASO. Significance: Simultaneously acquired fMRI/NIRS data processed using RIPTiDe allows us to isolate the contribution of hemodynamic fluctuations to the BOLD signal in every voxel. This permits significant reduction in the physiological noise in the BOLD data, and simultaneously yields an estimate of blood flow and volume at every location. This allows truly concurrent acquisition of high quality BOLD, CBV, and CBF information. Specific Aims: 1) Compare data quality obtained using NIRS from four different recording locations. The location of NIRS recording affects the purity of the measured hemodynamic signal. RIPTiDe images will be calculated using NIRS data from four probe locations, and compared to choose a standard recording location; 2) Evaluate the use of RIPTiDe data as input to the Balloon Model to generate quantitative estimates of blood flow and volume. We will use the RIPTiDe data (optimally delayed NIRS [HbR] and [tHb], and BOLD) as inputs to the balloon model, calculate CBV and CBF, and compare results and SNR/unit time with ASL and VASO; 3) Implement a RIPTiDe processing package. This aim will develop a streamlined data recording and analysis suite to simplify the use of RIPTiDe data, and allow other researchers to use this method. PUBLIC HEALTH RELEVANCE: The goal of this project is to improve existing multimodal processing of concurrently acquired NIRS and fMRI data to yield quantitative cerebral hemodynamic data (cerebral blood volume, cerebral blood flow, and mean transit time). The technique will be tested on 20 healthy subjects, and compared with arterial spin labeling and vascular space occupancy measurements.