Development and evaluation of an arterial spin-labeling digital reference object for quality control and comparison of data analysis applications
Development and evaluation of an arterial spin-labeling digital reference object for quality control and comparison of data analysis applications
复制标题
DOI：
10.1088/1361-6560/aaf83b
复制
发表时间：
2019-01-01
影响因子：
3.5
通讯作者：
Jackson, Edward F.
中科院分区：
文献类型：
--
作者：
Antolak, Alexander G.;Jackson, Edward F.
定量成像生物标志物（QIB）的纵向评估需要全面的质量控制（QC）计划，以最大限度地减少测量结果的偏倚和方差。此外，来自多个供应商的数据分析软件的可用性强调了对定量比较应用程序产生的计算QIB措施的手段的需要。这项工作的目的是描述一个数字参考对象（DRO），已开发的动脉自旋标记（ASL）测量结果的评价。ASL DRO是一个合成数据集，由10 x 10体素方块组成，具有一系列ASL对照图像信噪比（SNRControl）、血流（BF）和质子密度（PD）图像SNR值（SNRControl：1-100，BF：10-210 ml/100 g min（-1），SNRPD：10-100）。伪连续ASL序列用采集参数和根据通常与临床采集ASL图像相关的参数定义的建模信号强度进行模拟。使用市售nordicICE软件包（NordicNeuroLab，Inc，密尔沃基，WI）估计ASL参数。偏倚百分比测量和Bland-Altman分析表明，偏倚和方差随着SNR对照和BF值的增加而降低。对于高于5的SNRControl的所有BF值，观察到与参考值的极好一致性（对于所有SNRPD值，一致性相关系数大于0.92）。本工作中开发的ASL DRO允许评价具有生理意义的BF和SNRControl值之间的软件偏倚和方差。这些研究对于将基于ASL的定量BF测量转变为广泛的临床研究应用，并最终成为常规临床护理至关重要。
Longitudinal assessment of quantitative imaging biomarkers (QIBs) requires a comprehensive quality control (QC) program to minimize bias and variance in measurement results. In addition, the availability of data analysis software from multiple vendors emphasizes the need for a means of quantitatively comparing the computed QIB measures produced by the applications. The purpose of this work is to describe a digital reference object (DRO) that has been developed for the evaluation of arterial spin-labeling (ASL) measurement results. The ASL DRO is a synthetic data set consisting of 10 x 10 voxel square blocks with a range of ASL control image signal-to-noise ratio (SNRControl), blood flow (BF), and proton density (PD) image SNR values (SNRControl: 1-100, BF:10-210 ml/100 g min(-1), SNRPD:10-100). A pseudo-continuous ASL sequence was simulated with acquisition parameters and modeled signal intensities defined according to those typically associated with clinically-acquired ASL images. ASL parameters were estimated using the commercially-available nordicICE software package (NordicNeuroLab, Inc, Milwaukee, WI). Percent bias measures and Bland-Altman analyses demonstrated decreased bias and variance with increasing SNRControl and BF values. Excellent agreement with reference values was seen for all BF values above an SNRControl of 5 (concordance correlation coefficient greater than 0.92 for all SNRPD values). The ASL DRO developed in this work allows for the evaluation of software bias and variance across physiologically-meaningful BF and SNRControl values. Such studies are essential to the transition of quantitative ASL-based BF measurements into widespread clinical research applications, and ultimately, routine clinical care.