An adaptation of ridge regression for improved estimation of kinetic model parameters from PET studies
An adaptation of ridge regression for improved estimation of kinetic model parameters from PET studies
复制标题
DOI：
10.1109/tns.2004.843094
复制
发表时间：
2005-02-01
影响因子：
1.8
通讯作者：
Spence, AM
中科院分区：
文献类型：
--
作者：
Byrtek, M;O'Sullivan, F;Spence, AM
动态正电子发射断层扫描（PET）数据的定量分析，以获得动力学常数的房室模型涉及到使用非线性加权最小二乘回归。当前的估计技术通常具有差的均方误差估计特性。岭回归是一种技术，已被发现有潜力改善均方误差时，适用于非线性PET估计问题。然而，在这种情况下，岭回归的有效性，在很大程度上依赖于一个未知的偏置参数的正确选择和精确的规范的惩罚函数。在这项研究中，探讨了一种方法，提高岭回归的有效性，纳入更严格的贝叶斯公式的规格的岭罚函数。使用方差分量模型，先验协方差的岭惩罚项的开发。还评估了偏置参数的选择的自适应方法。偏置参数的自适应选择没有显示出比更标准的岭估计技术更好的估计。岭回归与贝叶斯制定的惩罚，但是，减少了当前的岭回归参数损失高达16%时，惩罚密切反映了真正的动力学参数协方差结构，并执行的惩罚时，目前的方法。
The quantitative analysis of dynamic positron emission tomography (PET) data to obtain kinetic constants in compartmental models involves the use of nonlinear weighted least squares regression. Current estimation techniques often have poor mean square error estimation properties. Ridge regression is a technique that has been found to have potential for improving mean square error when adapted to the nonlinear PET estimation problem. The effectiveness of ridge regression in this context, however, relies heavily on the correct selection of an unknown biasing parameter and the precise specification of a penalty function. In this study, an approach is explored for improving the effectiveness of ridge regression by incorporation of more rigorous Bayesian formulations for specification of the ridge penalty function. Using a variance component model, a prior covariance for the ridge penalty term is developed. An adaptive approach to the selection of the biasing parameter is also evaluated. The adaptive selection of the biasing parameter was not shown to improve estimation over more standard ridge estimation techniques. Ridge regression with the Bayesian formulation for the penalty, however, reduces current ridge regression parameter loss by up to 16% when the penalty closely reflects the true kinetic parameter covariance structure and performs comparably to the current method when the penalty does not.