Comparison study of temporal regularization methods for fully 5D reconstruction of cardiac gated dynamic SPECT

摘要

Temporal regularization plays a critical role in cardiac gated dynamic SPECT reconstruction, of which the goal is to obtain an image sequence from a single acquisition which shows simultaneously both cardiac motion and tracer distribution change over the course of imaging (termed 5D). In our recent work, we explored two different approaches for temporal regularization of the dynamic activities in gated dynamic reconstruction without the use of fast camera rotation: one is the dynamic EM (dEM) approach which is imposed on the temporal trend of the time activity of each voxel, and the other is a B-spline modeling approach in which the time activity is regulated by a set of B-spline basis functions. In this work, we extend the B-spline approach to fully 5D reconstruction, and conduct a thorough quantitative comparison with the dEM approach. In evaluation of the reconstruction results, we apply a number of quantitative measures on two major aspects of the reconstructed dynamic images: 1) accuracy of the reconstructed activity distribution in the myocardium, and 2) ability of the reconstructed dynamic activities to differentiate perfusion defects from normal myocardial wall uptake. These measures include: mean square error (MSE), bias-variance analysis (BV), accuracy of time activity curves (TAC), contrast- to-noise ratio (CNR) of a defect, composite kinetic map of the LV wall, and perfusion defect detectability with channelized Hotelling observer (CHO). In the experiments, we simulated gated cardiac imaging with the NURBS-based cardiac-torso (NCAT) phantom and Tc99m-Teboroxime as the imaging agent, where acquisition with the equivalent of only three full camera rotations was used during the imaging period. The results show that both dEM and B-spline 5D could achieve similar overall accuracy in the myocardium in terms of MSE. However, compared to dEM 5D, the B-spline approach could achieve a more accurate reconstruction of the voxel time-activity curves; in particular, B-spline 5D could achieve a much smaller bias level in the early uptake stage of the imaging period. Furthermore, it could allow better separation of the perfusion defect from the normal at both the early and late stages of the imaging period.