LMIRA: list-mode iterative reconstruction algorithm for SPECT

Abstract

The goal of this project is to investigate a new method for using list-mode data instead of binned projection data for single photon emission computed tomography (SPECT) reconstruction. List-mode acquisition was once standardly available for planar scintigraphy and is now becoming available for SPECT acquisition. The authors implemented maximum-likelihood iterative reconstruction with list-mode data for SPECT using a new projector/backprojector pair, which employs an intermediate stage to separate the source and camera specific components. In two dimensions, the intermediate stage is a circle surrounding the object. An emission radiance distribution is calculated at the circle from the estimate of the source distribution in the object. This leads to the probability that a photon, coming from the source distribution, leaves the circle surrounding the object at a certain point, going in a specific direction. The algorithm uses the emission radiance distribution to obtain the probabilities used in the iterative reconstruction method. The model takes into account distance-dependent resolution and nonuniform attenuation. Results illustrate how the individual steps of the algorithm work. The list-mode iterative reconstruction algorithm is compared to the standard MLEM iterative reconstruction method, which employs binned data. It is concluded that the finer sampling provided by list-mode data results in superior image quality. By using the intermediate stage, the authors obtained a structure that can very easily be adapted to different types of collimators and acquisition protocols.