Methodology Study Of Quantitive Analysis Of Positron Emission Tomography

In order to study the metabolism of glucose in human brain, the positron emission tomography (PET) is used, with the radiopharmaceutical18F-FDG as radiotracer, for the dynamic scanning of the brain. The metabolism of glucose in human brain can be described by a three compartment model. The parameters of the model can be estimated using the dynamic PET scanning data, and the cerebral metabolic rate of glucose can be figured out with these parameters. This paper primarily discussed the problems about the parameter estimation of the three compartment model using the FDG-PET data, including the following content:First, the quantitive analysis of three compartment model and the study of the influence factors is dicussed.At present, the methods of the quantitive analysis of the three compartment model includes the nonlinear least square, linear least square, general linear least square, Patlak graphical method and so on. These methods are tested using simulating data, and the advantages and disadvantages of them are analyzed. After that, several theoretical questions relating to the three compartment model are discussed. These questions include the uniqueness of parameters, the parameters of extended regions, the equilibrium of tracer between brain tissue and blood, the impact of the resolution of PET, the delayed scanning and using reference region to substituting the input function. Some of these questions are evaluated using simulating data.Second, the new method of derived input function is studyed. The quantitive analysis of PET data needs input and output function. The output function is the tissue time activity curve (TAC), which can be obtained through PET scanning data, and the input function is the TAC of the plasma TAC. In traditional method, the plasma TAC is usually acquired by amounts of arterious blood samples.In clinical application, the arterial sampling process is complicated and contains risks, which is the largest barrier for applying the quantitive analysis of PET data into clinical study. In this paper, two methods of deriving the input function in PET images are proposed based on the work of predecessors. Both of the two procedures are automatic, and blood samples for correcting the input fuction derived in images are two and one, respectively, thus the potential of applying them to clinical study is greatly increased. According to the characteristics of these methods, the method of no blood deriving is expectable when the performance of PET is improved.