Fast parametric image reconstruction for FDG using dynamic PET data

Measurement of the rate constant parameters of the tracer ¹⁸ F-FDG, used with positron emission tomography (PET) to determine the local cerebral metabolic rate of glucose (LCMRGlc), can provide a clear understanding of the physiological processes in the human brain. At present, the methods that are widely used to obtain LCMRGlc, such as nonlinear least squares (NLS), first require the reconstruction of a time sequence of images. The reconstruction of these images requires a large amount of computation, especially in 3D-PET, and the nonlinear based methods also require a large amount of computation. Here, the authors propose a fast parametric image reconstruction method for ¹⁸F-FDG dynamic PET studies. In their method a deconvolving process is first employed on the time sequential projection data to remove the effect of the measured plasma time activity. The deconvolved terms are integrated over three different time intervals and the parameters for determining LCMRGlc can be obtained analytically. The authors´ method requires only three reconstructing processes and reduces the computational demand to estimate LCMRGlc. The algorithm performance is evaluated using a digital phantom and a clinical data set and the results show that the proposed method produces images with the same or better quality as the images form the NLS method, with much less computation compared to the NLS method