Introduction: Image truncation in nuclear medicine is a common problem that can lead to artifacts in reconstructed images. We evaluate a modified single-photon emission computed tomography/computed tomography (SPECT/CT) acquisition and reconstruction method for truncated SPECT, which is guided by nontruncated CT. The method nearly eliminates truncation errors, and is ideal for cardiac imaging. We demonstrate its application on phantom and clinical cardiac SPECT/CT scans.
Methods: Tc-MIBI (2-methoxy isobutyl isonitrile) SPECT/CT scans were acquired on 14 patients, and on an anthropomorphic cardiac chest phantom. The original 34 x 34 cm field-of-view (FOV) projections were truncated to simulate a small 16 x 16 cm FOV acquisition. Data were reconstructed in three ways: (i) nontruncated and standard reconstruction (NTOSEM), which was our gold standard; (ii) truncated and standard reconstruction (TOSEM); and (iii) truncated and a modified reconstruction (TMOSEM). TMOSEM and TOSEM were both compared with NTOSEM by comparing relative count ratios in the heart, looking at the change in perfusion defect size, and comparing pixel correlation coefficients.
Results: Compared with NTOSEM, the use of TOSEM for small FOV clinical imaging incurred an average count ratio error greater than 100%, and decreased the calculated defect size by 17.13%. For TMOSEM, the average count ratio error was only 8.9%, and the defect size was only decreased by 0.19% compared with NTOSEM. When we plotted TOSEM against NTOSEM a correlation coefficient of 0.734 was calculated, and when we plotted TMOSEM against NTOSEM a correlation coefficient of 0.996 was measured. Comparing NTOSEM with TOSEM in the phantom study produced an average count ratio error greater than 100%. TMOSEM produced an error of 4.3% compared with NTOSEM.
Conclusion: Projection truncation due to small FOV cameras in cardiac SPECT/CT can lead to significant errors. TMOSEM guided by nontruncated CT reconstruction shows promise in reducing these errors.