Abstract
Dynamic PET imaging based on listmode data acquisition can provide timing information of coincidence events, which permit a flexible way to sort the data into several consecutive time frames. In this work, by using the listmode data combined with B-spline representation of radio-tracer time activity curve (TAC), we proposed a maximum likelihood expectation and maximization (MLEM) to estimate the parameters of TAC curve and restore the continuous TAC curve. Such a continuous TAC curve can provide better temporal resolution for dynamic PET imaging. In this pilot study, we evaluated the performance of the proposed method using clinical animal data acquired from an Inveon microPET scanner. We particularly focused on the FDG metabolic rate functions of heart and brain. Using these two organs as regions of interest (ROIs), the estimated TAC curves can depict the temporal variations of rate changes. In addition, the estimated TAC curves also matched with each other by using different sorting schemes (equal time and equal time). Since the listmode data can incorporate external signal inputs (like gating or respiratory signals), arbitrary temporal segmentation can be achieved as well. In future, we can combine such a generic segmentation into the proposed method and generate high-quality TAC curves corresponding to various physiological changes. The proposed method can be extended to 4D dynamic PET imaging which can make the best use of all collected events. It is optimistically anticipated that the proposed method can provide reliable radiotracer responses and will be useful for applications like drug development, therapy or diagnosis.