Maximum likelihood estimators for static and dynamic studies using emission tomography with auxiliary boundary information

Maximum likelihood (ML) estimation methods were developed for static and dynamic studies using emission tomography (ET) with auxiliary boundary information. The estimators are developed through parameterizing object boundaries and introducing registration parameters so that optimal uses of auxiliary boundary information can be practically implemented. The estimators are region-of-interest (ROI)-based ML estimators and are developed by implementing an observation model generalized from one that has been commonly used in iterative reconstruction. The ML estimator was used to simulated dynamic myocardial perfusion studies, compare simulation results with a previously established Cramer-Rao lower bound, and demonstrate that the ML estimator is close to the C-R bound in performance. Simulation results suggest that ET projection data are inherently capable of aligning auxiliary boundary information. The present model-based ML estimation methods provide an efficient way to extract position information from ET data.<>