Relationship between axial magnitude of respiratory motion in Tc-99m cardiac perfusion SPECT/CT stress acquisitions and reconstructed left ventricular short axis slice count density.

Objectives We have recently reported on respiratory motion compensation in 150+ Tc-99m SPECT/CT stress cardiac perfusion studies [SNMMI 2014, abstract 149] and concluded respiratory motion correction diminished motion artifacts and improved image fidelity with the extent of improvement related to the axial magnitude of estimated respiratory motion. Herein we study the relationship of the estimated axial respiratory motion, and the impact on left ventricular short axis slice count density using polar map quantification in a large group of clinically acquired patient studies.

Methods Patients undergoing routine Tc-99m Mibi same day rest-stress or two day stress-rest cardiac perfusion SPECT/CT tests were recruited for this study after institutional review board approval. The study group consisted of 775 patients (50.8% female) varying in age from 26 to 89 years with BMI’s varying from 19.1 to 66.2. Written consent was obtained from all patients. As before, patient motion was visually tracked using retro-reflective markers and infrared cameras (Vicon Motion Systems, Inc.) while acquiring in list mode employing a BrightView SPECT/CT (Philips Healthcare). A single marker tracked abdominal motion while others monitored possible rigid body motion. After synchronization, the visual tracking data were used to frame list mode data into 7 respiratory amplitude projections sets with a pixel size of 0.466 cm. Respiratory motion estimation was followed by iterative reconstruction combining the respiratory amplitude projection sets to a location between end in- and expiration to accommodate the averaging effect of respiratory motion on the cone-beam CT used for attenuation compensation. Data were also reconstructed without respiratory motion correction for comparison and all reconstructions included attenuation, resolution and scatter correction. Short axis slices were used to generate 17-segment polar maps after which %difference polar maps between RM corrected and no RM corrected data were obtained. Average increase in count density (n=775) and correlation between axial extent of RM estimates and the %difference polar map segments were calculated.

Results The table summarized the more significant average increases in count density and correlation values obtained from the comparison of the 17-segment polar maps. Increased count densities in the anterior, anterolateral, and inferior portions of the left ventricle correlated well with increasing axial (head-to-feet) respiratory motion estimates. Furthermore, the uniformity of the left ventricle was greatly improved when patients were deemed to have normal perfusion scans. Spatial resolution also improved visually. In our patient group, more than 30% recorded respiratory motion estimates larger than 1 cm, 10% recorded respiratory motion estimates larger than 1.4 cm, and 2.8% recorded respiratory motion estimates larger than 1.8 cm in the axial direction (head-to-feet) respectively, while less than 22% recorded respiratory motion estimates smaller than 0.6 cm. The average respiratory motion estimate in the axial direction was 0.89 cm.

Conclusions We showed the importance of respiratory motion compensation in a large number of clinically acquired cardiac perfusion SPECT/CT patient studies. Although patients with larger than 1 cm axial respiratory motion clearly benefitted, patients with less respiration improved as well.