Calculation of left ventricular volumes and ejection fraction from dynamic cardiac-gated (15)O-water PET/CT: 5D-PET

BACKGROUND: Quantitative measurement of myocardial blood flow (MBF) is of increasing interest in the clinical assessment of patients with suspected coronary artery disease (CAD). (15)O-water positron emission tomography (PET) is considered the gold standard for non-invasive MBF measurements. However, calculation of left ventricular (LV) volumes and ejection fraction (EF) is not possible from standard (15)O-water uptake images. The purpose of the present work was to investigate the possibility of calculating LV volumes and LVEF from cardiac-gated parametric blood volume (V (B)) (15)O-water images and from first pass (FP) images. Sixteen patients with mitral or aortic regurgitation underwent an eight-gate dynamic cardiac-gated (15)O-water PET/CT scan and cardiac MRI. V (B) and FP images were generated for each gate. Calculations of end-systolic volume (ESV), end-diastolic volume (EDV), stroke volume (SV) and LVEF were performed with automatic segmentation of V (B) and FP images, using commercially available software. LV volumes and LVEF were calculated with surface-, count-, and volume-based methods, and the results were compared with gold standard MRI. RESULTS: Using V (B) images, high correlations between PET and MRI ESV (r = 0.89, p < 0.001), EDV (r = 0.85, p < 0.001), SV (r = 0.74, p = 0.006) and LVEF (r = 0.72, p = 0.008) were found for the volume-based method. Correlations for FP images were slightly, but not significantly, lower than those for V (B) images when compared to MRI. Surface- and count-based methods showed no significant difference compared with the volume-based correlations with MRI. The volume-based method showed the best agreement with MRI with no significant difference on average for EDV and LVEF but with an overestimation of values for ESV (14%, p = 0.005) and SV (18%, p = 0.004) when using V (B) images. Using FP images, none of the parameters showed a significant difference from MRI. Inter-operator repeatability was excellent for all parameters (ICC > 0.86, p < 0.001). CONCLUSION: Calculation of LV volumes and LVEF from dynamic (15)O-water PET is feasible and shows good correlation with MRI. However, the analysis method is laborious, and future work is needed for more automation to make the method more easily applicable in a clinical setting.