Parameter estimation of cardiac geometry by ECG-gated PET imaging: validation using magnetic resonance imaging and echocardiography

Abstract

The purpose of this study was to apply and validate a previously developed model-based image analysis technique which derives estimates of regional myocardial wall thickness and the left ventricular radius directly from gated cardiac PET images.

Methods: In 11 normal volunteers, gated myocardial 18F-deoxyglucose (FDG) images with 16 equal gates spanning the entire cardiac cycle were acquired for 20 min. To improve count statistics and thus image quality, 3 and 5 of 16 gates were summed to obtain systolic and diastolic images. Based on a five-parameter model, radial profiles from systolic and diastolic PET images were fit by nonlinear regression for myocardial wall thickness, left ventricular radius and tracer activities in the blood pool, the myocardial tissue and the extracardiac background. Echocardiography and gated magnetic resonance imaging (MRI) were performed in 11 and 7 volunteers, respectively.

Results: We observed a significant (p < 0.001) correlation between measurements obtained by gated PET imaging and the correlative imaging modalities for myocardial wall thickness and left ventricular radius. While good agreement was observed between measurements of average radial shortening, estimates of average wall thickening differed significantly.

Conclusion: This model-based analysis offers accurate estimates of regional recovery coefficients directly from gated cardiac PET images and may also prove useful for the assessment of myocardial contractile function. These recovery coefficients are essential for the correction of partial volume effects when quantitative PET studies are performed.