Enhanced ejection fraction estimation in 4D respiratory motion compensated myocardial perfusion PET imaging

Objectives Respiratory (R) motion degrades myocardial perfusion (MP) PET imaging, affecting quantitative evaluation such as ejection fraction (EF) measurement. The goal is to study the effect of R motion and subsequent correction on measurement of left ventricular EF with normal and regionally reduced cardiac (C) motion.

Methods We developed a 4D R motion compensated image reconstruction method for dual-gated MP PET imaging. The R motion fields were first estimated from the R-gated only images with the end-expiration frame as the reference. For each C gate, the estimated R motion was then integrated within the 4D motion-compensated image reconstruction. Using the NCAT phantom, we simulated two dual-gated Rb-82 PET datasets, one with normal and the other with reduced C motion in the anterior-lateral region. Each dataset had 20 noise realizations of 5 R and 8 C gates, with the organ activity and noise levels representing clinically realistic Rb-82 scans. Using the Corrdior4DM software, we evaluated the EF from reconstructed C-gated images with and without R motion correction.

Results Both the estimated end-diastolic volume (EDV) and the end-systolic volume (ESV) were lower from images without than with R motion correction. In the normal C motion case, the underestimation was 26% for EDV and 14% for ESV. With and without R motion correction, the EF for the normal heart measured 60.6±1.3% and 54.6±2.3%, while the EF for the abnormal heart measured 52.3±1.8% and 49.4±2.8%. The difference between the average normal and the abnormal EFs was 8.3% with R motion correction compared to 4.7% without.

Conclusions R motion reduces the estimated EDV and ESV resulting in lowered EF measurements in both the normal and the abnormal C motion cases. Furthermore, R motion reduces the difference between estimated normal and abnormal EFs. The proposed 4D R motion correction may have the potential to improve clinical quantification and detection of abnormal EF.