Data-driven respiratory gating in cardiac PET/CT using the Positron Emission Tracking algorithm.

16 Objectives: PeTrack (Positron Emission Tracking) is a data-driven method for tracking positron emitting objects with in PET list-mode data. We evaluated its accuracy for respiratory motion tracking compared to that of the Real-Time Position Management (RPM) in cardiac PET/CT perfusion studies. Methods: This study included 51 rest and stress acquisitions of patients that underwent a rubidium-82 PET myocardial perfusion imaging study. Acquisitions were performed on a GE Discovery 690 scanner with list-mode and TOF capability. Respiratory monitoring was performed with the RPM system. A sodium-22 fiducial marker (92.3 kBq) was placed on each patient’s diaphragm for motion tracking with PeTrack. Respiratory triggers were extracted from the motion traces and used for respiratory-gated image reconstructions with six gates. Images were reconstructed using the vendor’s standard 3D ordered subset expectation maximization algorithm (3 iterations with 24 subsets) with CT-attenuation correction and correction for random and scatter coincidence events and post-reconstruction smoothing in the transaxial and axial directions. For each scan, the correlation coefficients between the two motion traces were measured as well as non-parametric descriptive statistics of the time intervals between respiratory triggers. From the reconstructed images, left ventricle wall thicknesses were estimated at the mid-HLA, mid-VLA and apical-SA regions. Cardiac motion between the end-inspiration and end-expiration respiratory phases along the axial direction was measured from each gated image set. Accurate gating was assumed to lead to reduced image blur, which lead to thinner LV wall thicknesses and increased inter-gate cardiac motion. All measured data were compared using the Wilcoxon signed-rank test. Results: The median correlation coefficient was R=0.46 with a 95% confidence interval of [0.01, 0.82] and 47 of the 51 scans were significantly correlated (p 0.05). The width of the respiratory interval distributions were estimated using the interquartile range, for which a statistically significant difference was observed (p = 0.04) indicating that the RPM system produced respiratory triggers with significantly less variation than PeTrack. Comparison of the reconstructed images indicated that there were no statistically significant differences (p > 0.05) for the myocardial wall thicknesses, in each region. The same was found for axial cardiac motion due to respiration measured in each set of images (p > 0.05). Conclusions: The PeTrack algorithm appeared to perform very similarly to the RPM system as a respiratory gating tool in cardiac PET/CT. The two methods provided respiratory signals that were generally well correlated and also exhibited comparable respiratory triggers for gated image reconstruction. The respiratory-gated images produced using PeTrack appeared to have no significant differences from those using the clinical RPM system. PeTrack appears to be an effective alternative to camera based respiratory gating systems in cardiac PET/CT and can be incorporated into the reconstruction workflow, removing the need for additional hardware.