Positron emission tomography (PET) images can be used to judge whether or not a person's bodily tissue is showing abnormal metabolism, providing a tool for early diagnosis and treatment of illnesses. Contemporary PET scanners have retracted their septa in order to increase the collected coincidental events. Thus, the protocol either needs to undergo three-dimensional image reconstruction, or use rebinning formulas to perform the less expensive two-dimensional image reconstruction for final images. Reconstruction using the second method saves image reconstruction time. The main goal of the paper is to further improve the performance by overlapping the rebinning and two-dimensional reconstruction operations, so as to early start in reconstruction, and to be able to undergo image reconstruction based on the pipelined direct sinograms. Frequency distance relations are analyzed in detail to generate the Fourier transformed sinograms in order for subsequent pipelined stages of reconstruction. The two-dimensional reconstruction operation does not have to wait until the completion of all sinogram generations, therefore it can hide most of the time spent in rebinning operations. The associated parameters can be pre-calculated indiscriminately beforehand for best performance.
Two-dimensional image reconstruction with spectrally-randomized ultrasound signals