Using the bootstrap method to evaluate image noise for investigation of axial collimation in hybrid PET

The purpose of this paper was to investigate the use of the bootstrap method to evaluate noise in three-dimensional (3-D) ordered-subsets expectation-maximization list-mode iterative reconstructions from data simulated on a hybrid positron emission tomography (PET) system. In addition, image noise from reconstructions acquired with various levels of axial collimation was investigated. A warm cylinder (20-cm diameter and 70-cm long) was simulated using the GEANT 3.21 Monte Carlo simulation code for different collimator designs: open-frame (OPEN), parallel (PAR) axial, and hybrid parallel fan-beam (HPF) axial collimators. Twenty noisy sets of list-mode data were simulated for each collimator. In addition, events from the first noisy list-mode data set of each category (OPEN, PAR axial, and HPF axial collimators) were chosen at random with replacement to produce twenty new sets of the same size. List-mode data were reconstructed using a 3-D ordered-subsets expectation-maximization (OSEM) list-mode iterative reconstruction algorithm using two subsets and up to 10 iterations. The standard deviations within volumes of interest (VOIs) were computed, and the mean value of this estimate over noise realizations was evaluated. The results indicate that there is no statistical significant difference in the standard deviation of VOI estimates from noisy realizations and bootstrap samples when using the OPEN, the PAR axial, and the HPF axial collimators. The results showed that the precision of VOI estimates varies with the axial spatial location within the object and the axial collimator. In addition, noise variation within list-mode reconstructed images appears to be related to the 3-D sensitivity map.