Myocardial perfusion quantification using simultaneously acquired ¹³NH₃-ammonia PET and dynamic contrast-enhanced MRI in patients at rest and stress

Purpose: Systematic differences with respect to myocardial perfusion quantification exist between DCE-MRI and PET. Using the potential of integrated PET/MRI, this study was conceived to compare perfusion quantification on the basis of simultaneously acquired ¹³NH₃-ammonia PET and DCE-MRI data in patients at rest and stress. Methods: Twenty-nine patients were examined on a 3T PET/MRI scanner. DCE-MRI was implemented in dual-sequence design and additional T₁ mapping for signal normalization. Four different deconvolution methods including a modified version of the Fermi technique were compared against ¹³NH₃-ammonia results. Results: Cohort-average flow comparison yielded higher resting flows for DCE-MRI than for PET and, therefore, significantly lower DCE-MRI perfusion ratios under the common assumption of equal arterial and tissue hematocrit. Absolute flow values were strongly correlated in both slice-average (R² = 0.82) and regional (R² = 0.7) evaluations. Different DCE-MRI deconvolution methods yielded similar flow result with exception of an unconstrained Fermi method exhibiting outliers at high flows when compared with PET. Conclusion: Thresholds for Ischemia classification may not be directly tradable between PET and MRI flow values. Differences in perfusion ratios between PET and DCE-MRI may be lifted by using stress/rest-specific hematocrit conversion. Proper physiological constraints are advised in model-constrained deconvolution.