Quantification of left ventricular volumes and ejection fraction in mice using PET, compared with MRI

PET has become an important noninvasive imaging technique in cardiovascular research for the characterization of mouse models in vivo. This modality offers unique insight into biochemical changes on a molecular level, with excellent sensitivity. However, morphologic and functional changes may be of equal importance for a thorough assessment of left ventricular (LV) pathophysiology. Although echocardiography and MRI are widely considered the imaging techniques of choice for the assessment of these parameters, their use with PET considerably increases study complexity and decreases cost- and time-efficiency. In this study, a novel method for the additional quantification of LV volumes and ejection fraction (EF) from PET was evaluated using cardiac MRI as the reference method.

The radiolabeled glucose derivative 18F-FDG was injected into 33 mice (6 mice with previous permanent occlusion of the left anterior descending artery [LAD], 15 mice with a temporary 30-min occlusion of the LAD, and 12 mice without previous surgery). 18F-FDG uptake within the LV myocardium was measured using a dedicated small-animal PET scanner. After we reconstructed the images into 16 electrocardiogram (ECG)-gated frames, we determined the LV cavity volumes in end-diastole (EDV) and end-systole (ESV) and the EF using a semiautomatic segmentation algorithm based on elastic surfaces. A 6.3-T cardiac MRI examination was performed in the same animals using an ECG-triggered and respiratory-gated multislice cine sequence. The MR images were segmented with a semiautomatic algorithm using commercially available software.

Overall, measurements from PET agreed well with those obtained by MRI. Mean EDV and ESV were slightly overestimated by PET (86+/-43 microL and 44+/-42 microL), compared with MRI (73+/-44 microL and 41+/-46 microL); mean (+/-SD) EF was similar (PET, 55+/-19 microL; MRI, 54+/-18 microL). Correlation between PET and MRI was excellent for EDV (0.97) and ESV (0.96) and good for EF (0.86). The slope of the regression line was nearly perfect for EDV (0.98) and EF (1.01) and slightly below 1 for ESV (0.90), indicating a good separation of abnormal and normal values with PET. The y-intercept was above zero for EDV (15 microL) and ESV (7 microL) and near to zero for EF (0.2%).

The quantification of LV volumes and EF in mice with PET is both efficient and accurate. This method allows for combined molecular and functional imaging of the left ventricle within a single scan, obviating additional sophisticated MRI in many cases.