First Investigation of List mode MLEM Reconstruction for Fast DC-SPECT System Design Optimization

A Dynamic Cardiac SPECT (DC-SPECT) system is being developed at the Massachusetts General Hospital, Harvard, featuring a cardio focus asymmetrical geometry to enable simultaneous high resolution and high sensitivity. We have generated 14 design iterations of the DC-SPECT with varying number of detector heads, system sensitivity, and system resolution. In such asymmetrical geometry the life cycle of design optimization from mechanical drawing to Monte Carlo implementation to image reconstruction and evaluation is long and labor intensive. This work presents a first implementation of List Mode Maximum Likelihood Expectation Maximization (LM-MLEM) reconstruction for the DC-SPECT system to enable fast design optimization using analytical method. We investigate the DC-SPECT system imaging resolution by applying a simple system modeling while implementing LM-MLEM. Results show that without any correction of scattering or attenuation effects, LM-MLEM reconstruction can achieve a 5 mm FWHM spatial resolution. The results agree with the expected performance of the DC-SPECT. Specifically, we show the advantages of analytic based system matrix model over Monte-Carlo (MC) based models. A comparison between images reconstructed by analytic modeling and MC methods is presented.