Graham M M
Department of Radiology, University of Washington, Seattle 98195-6113, USA.
J Nucl Med. 1997 Jul;38(7):1161-8.
Blood or plasma time-activity curves (TACs) are used as the input function for mathematical models of tracer kinetics in several applications including PET. Uncertainty associated with both the blood data and the PET tissue data can result in uncertainty in the estimates of metabolic rates, blood flow, etc.
This article presents an approach to reduce the uncertainty in the blood TAC by fitting a model to the curve. The model includes a choice of bolus or infusion input and has three compartments (plasma, interstitial fluid and tissue fluid) with exchange between them. There is a parameter for loss from the plasma compartment. To test the utility of smoothing blood TACs with this approach, a program was set up, using the fluorodeoxyglucose (FDG) model, with simulated noisy blood and tissue TACs. The smoothed blood TAC was compared to a linearly interpolated TAC as the input function with a compartmental model parameter estimation program and with graphical analysis.
With a well sampled blood TAC (19 points), the model approach is somewhat more accurate than linear interpolation if the s.d. of noise added to the data exceeded 10%. With sparsely sampled blood TACs (five points) or with a large gap in the blood TAC, the modeled approach was markedly better. For graphical analysis, the model smoothed TAC was also more accurate, although, in general, the results were not as sensitive to the input function.
This approach, using a physiologically reasonable model to smooth the blood TAC, is a useful aid in PET data analysis, particularly when the data are quite noisy or when there are large gaps in the data.
在包括正电子发射断层扫描（PET）在内的多种应用中，血液或血浆时间 - 活性曲线（TAC）被用作示踪剂动力学数学模型的输入函数。血液数据和PET组织数据的不确定性都可能导致代谢率、血流量等估计值的不确定性。
本文提出了一种通过对曲线拟合模型来降低血液TAC不确定性的方法。该模型包括推注或输注输入的选择，并有三个隔室（血浆、间质液和组织液），它们之间存在交换。血浆隔室有一个损失参数。为了测试用这种方法平滑血液TAC的效用，使用氟脱氧葡萄糖（FDG）模型，设置了一个程序，使用模拟的有噪声的血液和组织TAC。将平滑后的血液TAC与线性插值TAC作为输入函数，通过隔室模型参数估计程序和图形分析进行比较。
对于采样良好的血液TAC（19个点），如果添加到数据中的噪声标准差超过10%，模型方法比线性插值更准确。对于采样稀疏的血液TAC（五个点）或血液TAC中有大的间隙时，建模方法明显更好。对于图形分析，模型平滑后的TAC也更准确，尽管一般来说，结果对输入函数的敏感度较低。
这种使用生理上合理的模型来平滑血液TAC的方法，在PET数据分析中是一种有用的辅助手段，特别是当数据噪声很大或数据中有大的间隙时。
