Comparison of volumetric and functional parameters in simultaneous cardiac PET/MR: feasibility of volumetric assessment with residual activity from prior PET/CT

Abstract

Objective: To compare cardiac left ventricular (LV) parameters in simultaneously acquired hybrid fluorine-18-fluorodeoxyglucose ([18F] FDG) positron emission tomography/magnetic resonance imaging (PET/MRI) in patients with residual tracer activity of upstream PET/CT. Methods: Twenty-nine patients (23 men, age 58±17 years) underwent cardiac PET/MRI either directly after a non-cardiac PET/CT with homogenous cardiac [18F] FDG uptake (n=20) or for viability assessment (n=9). Gated cardiac [18F] FDG PET and cine MR sequences were acquired simultaneously and evaluated blinded to the cross-imaging results. Image quality (IQ), end-diastolic (LVEDV), end-systolic volume (LVESV), ejection fraction (LVEF) and myocardial mass (LVMM) were measured. Pearson correlation and intraclass correlation coefficient (ICC), regression and a Bland–Altman analysis were assessed. Results: Except LVMM, volumetric and functional LV parameters demonstrated high correlations (LVESV: r=0.97, LVEDV: r=0.95, LVEF: r=0.91, LVMM: r=0.87, each p<0.05), but wide limits of agreement (LOA) for LVEDV (−25.3–82.5ml); LVESV (−33.1–72.7ml); LVEF (−18.9–14.8%) and LVMM (−78.2–43.2g). Intra- and interobserver reliability were very high (ICC≥0.95) for all parameters, except for MR-LVEF (ICC=0.87). PET-IQ (0–3) was high (mean: 2.2±0.9) with significant influence on LVMM calculations only. Conclusion: In simultaneously acquired cardiac PET/MRI data, LVEDV, LVESV and LVEF show good agreement. However, the agreement seems to be limited if cardiac PET/MRI follows PET/CT and only the residual activity is used. Key Points • [ 18 F] FDG PET-MRI is feasible with residual [ 18 F] FDG activity in patients with homogenous cardiac uptake. • Cardiac volumes and function assessed by PET/MRI show good agreement. • LVEDV and LVESV are underestimated; PET overestimates LVMM and LVEF. • Cardiac PET and MRI data correlate better when acquired simultaneously than sequentially. • PET and MRI should not assess LV parameters interchangeably.