Fully 5D reconstruction of gated dynamic cardiac SPECT images

In our previous work we proposed a unified image reconstruction approach for dynamic images from traditional gated SPECT acquisition. In this approach we divide the cardiac cycle into a number of gate intervals as in gated SPECT, but treat the tracer distribution for each gate as a time-varying signal. In this work we extend this approach to fully five-dimensional (5D) image reconstruction, which can produce a volumetric dynamic image sequence that shows both cardiac motion and time-varying tracer distribution simultaneously. We simulated gated cardiac perfusion imaging using the NURBS-based cardiac-torso (NCAT) phantom with Tc99m-Teboroxime as the imaging agent. Our results show that use of motion-compensated temporal regularization in the proposed approach can lead to much improved reconstruction than using spatial smoothness alone.