Effects of acquisition time and reconstruction algorithm on image quality, quantitative parameters, and clinical interpretation of myocardial perfusion imaging.

BACKGROUND: Recently introduced iterative reconstruction algorithms with resolution recovery (RR) and noise-reduction technology seem promising for reducing scan time or radiation dose without loss of image quality. However, the relative effects of reduced acquisition time and reconstruction software have not previously been reported. The aim of the present study was to investigate the influence of reduced acquisition time and reconstruction software on quantitative and qualitative myocardial perfusion single photon emission computed tomography (SPECT) parameters using full time (FT) and half time (HT) protocols and Evolution for Cardiac Software.
METHODS: We studied 45 consecutive, non-selected patients referred for a clinically indicated routine 2-day stress/rest (99m)Tc-Sestamibi myocardial perfusion SPECT. All patients underwent an FT and an HT scan. Both FT and HT scans were processed according to our standard procedure with both ordered-subset expectation maximization (OSEM) + filtered back projection (FBP) reconstructions and a second reconstruction of HT scans was performed with the RR software producing three datasets for each patient for visual analysis (FT-OSEM, HT-OSEM, and HT-RR) and for quantitative analysis (FT-FBP, HT-FBP, and HT-RR). The datasets were analyzed using commercially available QGS/QPS software and read by two observers evaluating image quality and clinical interpretation. Image quality was assessed on a 10-cm visual analog scale score.
RESULTS: HT imaging was associated with loss of image quality that was compensated for by RR reconstruction. HT imaging was also associated with increasing perfusion defect extents, an effect more pronounced using RR than FBP reconstruction. Compared to standard FT-FBP, HT-RR significantly reduced left ventricular volumes whereas HT-FBP increased end-systolic volume. HT imaging had no effect on measured left ventricular ejections fraction or measures of reversibility. Image interpretation found a higher level of concordance between FT-OSEM and HT-RR than between FT-OSEM and HT-OSEM without any observable systematic effects.
CONCLUSIONS: Use of RR reconstruction algorithms compensates for loss of image quality associated with reduced scan time. Both HT acquisition and RR reconstruction algorithm had significant effects on motion and perfusion parameters obtained with standard software, but these effects were relatively small and probably of limited clinical importance. Although no systematic effects on image interpretation were observed, the influence on diagnostic accuracy remains to be determined.