Hopper J L, Davis S M, Tress B M, Kaye A H, Rossiter S C, Derrick P L
University of Melbourne, Faculty of Medicine Epidemiology Unit, Australia.
Invest Radiol. 1987 Aug;22(8):651-7. doi: 10.1097/00004424-198708000-00006.
Dynamic computed tomography (DCT) of the brain can be used to study the transit time of first passage of a bolus injection of intravenous contrast medium. Comparison of cerebral perfusion with corresponding sites in the left and right cerebral hemispheres is of diagnostic interest because a real difference may be indicative of differential damage. A method for estimating the mean transit time and approximating its standard error, by assuming a gamma function for the response curve and an appropriate error structure, is presented. Expressed as log-linear regression, estimation is achieved by maximum likelihood using a statistical package such as GLIM or SPSSX. Statistical comparison of the mean transit time to or between corresponding sites can be made; issues of model fit and biologic interpretation need to be considered as an integral part of statistical inference. These methods enable users of CT equipment (without specific software for estimation of mean transit time) to use any log-linear routine for diagnostic purposes. An example of the fit procedure and interpretation in the light of clinical evidence is given.
脑部动态计算机断层扫描（DCT）可用于研究静脉注射造影剂团注首次通过的传输时间。比较左右脑半球相应部位的脑灌注情况具有诊断意义，因为实际差异可能表明存在不同程度的损伤。本文提出了一种通过假设响应曲线为伽马函数并采用适当误差结构来估计平均传输时间及其标准误差近似值的方法。以对数线性回归表示，使用GLIM或SPSSX等统计软件包通过最大似然法进行估计。可以对平均传输时间与相应部位之间或相互之间进行统计比较；模型拟合和生物学解释问题应作为统计推断的一个组成部分加以考虑。这些方法使CT设备用户（无需用于估计平均传输时间的特定软件）能够将任何对数线性程序用于诊断目的。文中给出了根据临床证据进行拟合程序和解释的示例。
