Nakamura M, Suzuki Y, Kobayashi S
IEEE Trans Biomed Eng. 1989 Sep;36(9):906-17. doi: 10.1109/10.35299.
A method based on the maximum entropy principle is described for recovering physiological components from a sequence of dynamic radionuclide images. The method does not require any assumption about the shapes of the component curves. It is, however, assumed that the components are nonnegative. The method requires normalization of component curves. The sum of squares of each component curve to be estimated is normalized so as to be unity. A first-pass cardiac study and a hepatic study are selected for investigation of applicability of the method. The method is investigated numerically using computer-generated time-activity curves with six components for the cardiac study and seven components for the hepatic study. Within the regions studied, the results have demonstrated that the method may recover the component time-activity curves and the corresponding images of physiological components from a sequence of dynamic radionuclide images. The ability and limitations of the method to recover physiological components are discussed and potential applications are suggested.
描述了一种基于最大熵原理的方法，用于从动态放射性核素图像序列中恢复生理成分。该方法不需要对成分曲线的形状做任何假设。然而，假定成分是非负的。该方法需要对成分曲线进行归一化。待估计的每条成分曲线的平方和被归一化至1。选择首次通过心脏研究和肝脏研究来考察该方法的适用性。使用计算机生成的时间-活度曲线对该方法进行数值研究，心脏研究中有六个成分，肝脏研究中有七个成分。在所研究的区域内，结果表明该方法可以从动态放射性核素图像序列中恢复成分时间-活度曲线以及相应的生理成分图像。讨论了该方法恢复生理成分的能力和局限性，并提出了潜在的应用。
