Dual-motion frozen reconstruction technique improves the diagnostic accuracy of myocardial perfusion SPECT imaging in CZT cameras

Objectives Myocardium is constantly moving during image acquisition. The motions can be categorized into centroid shift (respiration) and deformation (myocardial contraction). We proposed the dual-motion frozen reconstruction (DMFR) technique in cadmium-zinc-telluride (CZT) cameras, which may allow better noise modeling than the traditional morphing and summing technique. We also assessed whether the effect improves the diagnostic accuracy of myocardial perfusion imaging (MPI) in the detection of coronary artery disease (CAD).

Methods A total of 82 consecutive patients who had stable angina without documented myocardial infarction or coronary revascularization were evaluated. Treadmill-stress/rest Tl-201 MPI with CZT cameras and then invasive coronary angiography were performed within 6 months. Respiratory and cardiac-gated (5 and 8 bins, respectively) images were generated using the acquired list file. The non-rigid deformation fields (DFs) were calculated by optical flow algorithm. With these DFs, 40 respiratory-cardiac gated projections were reconstructed into 1 motion frozen SPECT image using DMFR technique. Semi-quantitative analysis with five-point score on 17-segment model was applied on the uncorrected and the motion-frozen images. Their difference was compared with receiver operating characteristic (ROC) curve analysis using ≥70% coronary luminal stenosis as cutoff.

Results More respiratory motion in the post-stress scans was noted (stress/rest=10±3/8±3 mm, P<0.001). Compared to uncorrected images, the study showed significantly improved diagnostic accuracy for DMFR images in patient-based analysis (AUC=0.926 vs. 0.828, P=0.001), as well as LAD and RCA (0.934 vs. 0.863, P=0.002; 0.936 vs. 0.842, P=0.017), but not LCX (0.938 vs. 0.910, P=0.286), in vessel-based analysis.

Conclusions The DMFR technique in CZT cameras significantly improves the diagnostic accuracy of treadmill-stressed MPI in the detection of CAD, especially in LAD and RCA vascular territories.