Ph.D. Thesis Defense by Ji Chen 
Monday, November 11, 2002

(Drs. John Valentine and James R. Galt, co-advisors)

"Quantitative Cardiac SPECT in the Presence of Changing Radionuclide Distributions"


Conventional myocardial perfusion single photon emission computed tomography (SPECT) is based on static tracer distribution. In 99mTc teboroxime SPECT, however, the tracer washes out from the myocardium and washes into the liver during the data acquisition. Reconstructions from the inconsistent projections may contain significant artifacts if conventional SPECT acquisition and reconstruction techniques are used. High liver uptake can result in overlaps between the liver and myocardium and make it difficult to diagnose the inferior wall.

A computer model of 99mTc teboroxime SPECT acquisition has been developed. Photon attenuation, collimator resolution, Compton scatter, Poisson noise, and tracer kinetics are included in the simulation. Novel acquisition and processing techniques can be more easily tested with the simulation than in phantom and patient studies.

The fast fanning acquisition protocol has been proposed to optimize 99mTc teboroxime SPECT. It can produce a series of dynamic scans, from which the trace kinetics can be extracted by a region of interest method. Summing these dynamic scans and least-squares restoration (LSR) can minimize artifacts due to the tracer washout, whereas nonlinear regression with a distance penalty (NRDP) can remove the liver contamination from the myocardium. Based on these three methods, a processing protocol has been developed for optimization of 99mTc teboroxime SPECT. This protocol consists of three steps: preprocessing (Summing or LSR), reconstruction, and postprocessing (NRDP). This protocol has been tested with simulation studies in patient and heart motion, which can effectively produce changing activity distributions. The changing tracer distribution induced by patient and heart motion may produce artifacts in the reconstructions, but it has little impact on the processing protocol. Motion correction, if available, could be applied to the reconstructions before the postprocessing step in the protocol, potentially producing images of improved quality.