Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Medical Biophysics


Robert Z. Stodilka


Regenerative therapy via stem cell transplantation has received increased attention to help treat the myocardial injury associated with heart disease. Currently, the hybridisation of SPECT with X-ray CT is expanding the utility of SPECT. This thesis compared two SPECT/CT systems for attenuation correction using slow or fast-CT attenuation maps (mu-maps). We then developed a method to localize transplanted cells in relation to compromised blood flow in the myocardium following a myocardial infarction using SPECT/CT. Finally, a method to correct for image truncation was studied for a new SPECT/CT design that incorporated small field-of-view (FOV) detectors. Computer simulations compared gated-SPECT reconstructions using slow-CT and fast-CT mu-maps with gated-CT mu-maps. Using fast-CT mu-maps improved the Root Mean Squared (RMS) error from 4.2% to 4.0%. Three canine experiments were performed comparing SPECT/CT reconstruction using the Infinia/Hawkeye-4 (slow-CT) and Symbia T6 (fast-CT). Canines were euthanized prior to imaging, and then ventilated. The results showed improvements in both RMS errors and correlation coefficients for all canines. A first-pass contrast CT imaging technique can identify regions of myocardial infarction and can be fused with SPECT. Ten canines underwent surgical ligation of the left-anterior-descending artery. Cells were labeled with 111In-tropolone and transplanted into the myocardium. SPECT/CT was performed on day of transplantation, 4, and 10 days post-transplantation. For each imaging session first-pass perfusion CT was performed and successfully delineated the infarct zone. Delayed-enhanced MRI was performed and correlated well with first-pass CT. Contrast-to-noise ratios were calculated for 111In-SPECT and suggested that cells can be followed for 11 effective half-lives. We evaluated a modified SPECT/CT acquisition and reconstruction method for truncated SPECT. Cardiac SPECT/CT scans were acquired in 14 patients. The original projections were truncated to simulate a small FOV acquisition. Data was reconstructed in three ways: non-truncated and standard reconstruction (NTOSEM), which was our gold-standard; truncated and standard reconstruction (TOSEM); and truncated and a modified reconstruction (TMOSEM). Compared with NTOSEM, small FOV imaging incurred an average cardiac count ratio error greater than 100% using TOSEM and 8.9% using TMOSEM. When we plotted NTOSEM against TOSEM and TMOSEM the correlation coefficient was 0.734 and 0.996 respectively.