Doury Maxime, Dizeux Alexandre, de Cesare Alain, Lucidarme Olivier, Pellot-Barakat Claire, Bridal S Lori, Frouin Frédérique
Laboratoire d'Imagerie Biomédicale (LIB), CNRS, Inserm, UPMC Univ. Paris 06, Sorbonne Universités, Paris, France.
Phys Med Biol. 2017 Feb 7;62(3):1113-1125. doi: 10.1088/1361-6560/aa54a3. Epub 2016 Dec 19.
Dynamic contrast-enhanced ultrasound has been proposed to monitor tumor therapy, as a complement to volume measurements. To assess the variability of perfusion parameters in ideal conditions, four consecutive test-retest studies were acquired in a mouse tumor model, using controlled injections. The impact of mathematical modeling on parameter variability was then investigated. Coefficients of variation (CV) of tissue blood volume (BV) and tissue blood flow (BF) based-parameters were estimated inside 32 sub-regions of the tumors, comparing the log-normal (LN) model with a one-compartment model fed by an arterial input function (AIF) and improved by the introduction of a time delay parameter. Relative perfusion parameters were also estimated by normalization of the LN parameters and normalization of the one-compartment parameters estimated with the AIF, using a reference tissue (RT) region. A direct estimation (rRTd) of relative parameters, based on the one-compartment model without using the AIF, was also obtained by using the kinetics inside the RT region. Results of test-retest studies show that absolute regional parameters have high CV, whatever the approach, with median values of about 30% for BV, and 40% for BF. The positive impact of normalization was established, showing a coherent estimation of relative parameters, with reduced CV (about 20% for BV and 30% for BF using the rRTd approach). These values were significantly lower (p < 0.05) than the CV of absolute parameters. The rRTd approach provided the smallest CV and should be preferred for estimating relative perfusion parameters.
动态对比增强超声已被提议用于监测肿瘤治疗，作为体积测量的补充手段。为了评估在理想条件下灌注参数的变异性，在小鼠肿瘤模型中进行了四项连续的重测研究，采用控制注射的方式。随后研究了数学建模对参数变异性的影响。在肿瘤的32个子区域内估计基于组织血容量（BV）和组织血流（BF）的参数的变异系数（CV），将对数正态（LN）模型与由动脉输入函数（AIF）馈入并通过引入时间延迟参数进行改进的单室模型进行比较。还通过使用参考组织（RT）区域对LN参数进行归一化以及对用AIF估计的单室参数进行归一化来估计相对灌注参数。基于不使用AIF的单室模型，通过使用RT区域内的动力学，也获得了相对参数的直接估计值（rRTd）。重测研究结果表明，无论采用何种方法，绝对区域参数的CV都很高，BV的中位数约为30%，BF的中位数约为40%。确定了归一化的积极影响，显示相对参数的估计具有一致性，CV降低（使用rRTd方法时，BV约为20%，BF约为30%）。这些值显著低于（p < 0.05）绝对参数的CV。rRTd方法提供了最小的CV，在估计相对灌注参数时应优先选用。
