Elastic motion correction improves Rb-82 Cardiac PET ECG-gated image quality

Objectives: ECG-gated imaging for evaluation of left ventricular (LV) contractile function is an integral part of myocardial perfusion imaging (MPI) but can be challenging with Rb-82 due to the short half-life and low count-statistics. Motion-compensated image reconstruction was reported recently on some latest-generation PET-CT scanners, but has not been optimized for Rb-82 PET MPI. This study evaluated the effect of elastic motion-compensated (MOCO) image reconstruction on ECG-gated Rb-82 PET image quality.

Methods: ECG-gated and ungated (static) images were analyzed at rest and stress from N=20 sequential patients referred for Rb-82 MPI (9 MBq/kg) on a PET-CT scanner with ≍200 ps time-of-flight (TOF) resolution. Standard (no-MOCO) ECG-gated images were reconstructed using 6 mm Gaussian filter, and used to estimate contractile and respiratory motion vector fields (MVF). Additional ECG-gated images were then reconstructed using the contractile-MVF (single-MOCO) and combined respiratory- and contractile-MVF (dual-MOCO) information integrated into the iterative reconstruction algorithm (OSEM with 4 iterations and 5 subsets). Static (ungated) images at rest were also reconstructed using 2, 4, 6 mm Gaussian filters for comparison of image quality. Myocardium signal recovery was measured as the maximum activity in the left ventricle (LV) at end-diastole (ED). Background signal and noise were measured as the left atrium blood cavity mean and standard deviation, also at the ED phase. LV myocardium signal-to-noise ratio (SNR) and myocardium-to-blood contrast-to-noise ratio (CNR) values were calculated for the static and ECG-gated images. SNR and CNR were compared between reconstruction methods using paired t-tests.

Results: End-diastolic image SNR and CNR increased in 95% (or 55%) of patients at rest using single-MOCO (or dual-MOCO) compared to standard ECG-gated reconstruction. Similarly at stress, SNR and CNR increased in 100% (or 60%) of patients using single-MOCO (or dual-MOCO) reconstruction. Single-MOCO reconstruction significantly improved SNR (+48%) and CNR (+51%), both at stress (+43%) and rest (+56%) (all P < 0.0001), as shown in the figures below. By contrast, dual-MOCO reconstruction showed only modest benefits in SNR (+9%) and CNR (+15%), at stress (+8%) or rest (+15%) on average (0.13 > P > 0.01), with 40% of scans resulting in lower image quality compared to the standard (no-MOCO) gated reconstruction. The dual-MOCO gated images reconstructed with 6mm filter had image SNR and CNR that were similar to static ungated image reconstructed with 2 mm filtering, whereas the single-MOCO gated images were similar to the static images with 4 mm filtering. Both single- and dual-MOCO images had lower SNR and CNR compared to the static images reconstructed with the same 6 mm filter, suggesting that there was some residual noise in the single-MOCO estimated MVF that could be further improved. Image quality (SNR and CNR) decreased with patient weight (both P < 0.05), likely due to increased attenuation effects despite the use of proportional weight-based dosing, suggesting that larger patients require even higher administered activity to achieve uniform image quality.

Conclusions: Single (contractile) motion-compensated image reconstruction improved the end-diastolic image SNR and CNR over standard uncompensated or dual motion-compensated image reconstruction, and is recommended for optimal ECG-gated image quality using Rb-82 on a current-generation TOF PET-CT scanner.

• Download figure
• Open in new tab
• Download powerpoint