Model-based analysis of electrocardiography-gated cardiac (18)F-FDG PET images to assess left ventricular geometry and contractile function.

UNLABELLED This study presents and evaluates a model-based image analysis method to calculate from gated cardiac (18)F-FDG PET images diastolic and systolic volumes, ejection fraction, and myocardial mass of the left ventricle. The accuracy of these estimates was delineated using measurements obtained by MRI, which was considered the reference standard because of its high spatial resolution. METHODS Twenty patients (18 men, 2 women; mean age +/- SD, 59 +/- 12 y) underwent electrocardiography-gated cardiac PET and MRI to acquire a set of systolic and diastolic short-axis images covering the heart from apex to base. For PET images, left ventricular radius and wall thickness were estimated by model-based nonlinear regression analysis applied to the observed tracer concentration along radial rays. Endocardial and epicardial contours were derived from these estimates, and left ventricular volumes, ejection fraction, and myocardial mass were calculated. For MR images, an expert manually drew contours. RESULTS Left ventricular volumes by PET and MRI were 101 +/- 60 mL and 112 +/- 93 mL, respectively, for end-systolic volume and 170 +/- 68 mL and 189 +/- 99 mL, respectively, for end-diastolic volume. Ejection fraction was 44% +/- 13% by PET and 46% +/- 18% by MRI. The left ventricular mass by PET and MRI was 196 +/- 44 g and 200 +/- 46 g, respectively. PET and MRI measurements were not statistically significant. A significant correlation was observed between PET and MRI for calculation of end-systolic volumes (r = 0.93, SEE = 23.4, P < 0.0001), end-diastolic volumes (r = 0.92, SEE = 26.7, P < 0.0001), ejection fraction (r = 0.85, SEE = 7.4, P < 0.0001), and left ventricular mass (r = 0.75, SEE = 29.6, P < 0.001). CONCLUSION Model-based analysis of gated cardiac PET images permits an accurate assessment of left ventricular volumes, ejection fraction, and myocardial mass. Cardiac PET may thus offer a near-simultaneous assessment of myocardial perfusion, metabolism, and contractile function.