A study of the influence of local variations in myocardial thickness on SPECT perfusion imaging

The objective of this project was to study the impact of local variations in myocardial thickness on single photon emission computed tomography (SPECT) perfusion imaging using segmented high-resolution gated-computer tomography (CT) data of the heart to define source distributions. High-resolution breath-hold gated CT data of 16 patients were obtained using a GE Lightspeed multislice CT system. Transverse slices (512 × 512) with x-y pixel size of 0.039 cm and a slice thickness of either 0.25 cm or 0.125 cm were segmented to obtain left-ventricular (LV) and right-ventricular (RV) myocardial walls for all time frames of the cardiac cycle. Polar maps of myocardial thickness were generated using radial and short-axis slices of the LV. Absolute quantitation was performed on the myocardium thickness polar maps. The segmented LV and RV were also incorporated in the gated MCAT phantom, and analytical SPECT projections generated using typical acquisition parameters. Projection data were folded down to 64 × 64 image matrices and reconstructed using FBP without attenuation correction (FBP_NC), five iterations of ordered-subset expectation-maximization (OSEM) including attenuation compensation (OSEM_AC), as well as five iterations of OSEM including AC and compensation for detector response (OSEM_ACR). CEQUAL polar maps were generated and relative quantitative analysis performed. Results show there is considerable variation in wall thickness at the resolution of the CT studies. Apical thinning, papillary muscles, and apical to base shortening are clearly visible. OSEM_AC and OSEM_ACR give a more realistic perfusion distribution in the simulated MCAT patient data than FBP_NC.