Quantitative Evaluation Of Pet Image Using Event Information Bootstrap

The purpose of this study was to enhance the effect in the PET image quality according to event bootstrap of small animal PET data. In order to investigate the time difference condition, realigned sinograms were generated from randomly sampled data set using bootstrap. List-mode data was obtained from small animal PET scanner for Ge-68 30 sec, Y-90 20 min and Y-90 60 min. PET image was reconstructed by Ordered Subset Expectation Maximization(OSEM) 2D with the list-mode format. Image analysis was investigated by Signal to Noise Ratio(SNR) of Ge-68 and Y-90 image. Non-parametric resampled PET image SNR percent change for the Ge-68 30 sec, Y-90 60 min, and Y-90 20 min was 1.69 %, 7.03 %, and 4.78 %, respectively. SNR percent change of non-parametric resampled PET image with time difference condition was 1.08 % for the Ge-68 30 sec, 6.74 % for the Y-90 60 min and 10.94 % for the Y-90 29 min. The result indicated that the bootstrap with time difference condition had a potential to improve a noisy Y-90 PET image quality. This method should be expected to reduce Y-90 PET measurement time and to enhance its accuracy.