Identifying and correcting organ settling motion during Tc-99m cardiac SPECT/CT perfusion imaging.

Objectives: Upward creep of the heart during exercise Tl-201 cardiac perfusion SPECT imaging is well described^1-2, however, a similar phenomenon after delayed Tc-99m cardiac perfusion SPECT imaging where patients are well rested has not been given as much attention³. In our recent research^4-6, developing rigid body and respiratory motion compensation techniques employing a visual tracking system, we observed a discrepancy in some patients between rigid body motion in the anterior-posterior direction tracked by retro-reflective markers placed on the chest and a single retro-reflective marker placed on the abdomen after removing the influence of respiration. The objective of this study was to retrospectively investigate the number of patients with and the effects of the abovementioned discrepancy in a large number of cardiac perfusion SPECT/CT patients with visual tracking available.

Methods: As in our previous work^4-6, 1103 patients (556 female) with Institutional Review Board (IRB) approved written consent were acquired in list mode on a BrightView SPECT/CT (Philips Healthcare, Cleveland OH) in conjunction with visual tracking using retro-reflective markers (Vicon Motion Systems, Inc., Lake Forest, CA). Information of the imaging bed’s motion was used to synchronize the two modalities. A 3 degrees of freedom (3DOF) rigid body motion estimate (translational motion only) was additionally obtained from the abdominal retro-reflective marker and the anterior-posterior component added to the 6 DOF rigid body motion head-to-feet component. The resulting rigid body motion estimates were combined with the respiratory motion estimates. Patients were divided into two categories, without or with measurable anterior-posterior motion according to the 3DOF estimate. Using an ordered-subsets expectation-maximization (OSEM) iterative reconstruction algorithm employing a 3-dimensional Gaussian rotator, it was possible to compensate for the discrepancy in all patients identified. For comparison, projection data were also reconstructed without compensating for the additional motion observed. All other physical degradations (attenuation, Compton scatter, and distance dependent resolution) were also accounted for during reconstruction. Visual evaluation and polar map quantitation were employed to evaluate changes in the reconstructed slices.

Results: Six hundred and twenty-three (623) patients (298 female) were identified as having additional abdominal motion (4.61 mm ± 2.58 mm, range 1.1 mm - 26.60 mm), with motion recorded by females statistically significantly less than males (p=1.83E-18). The visual improvement (shape and count uniformity) and count density changes were more pronounced for larger motion estimates.

Conclusion: We visually successfully corrected the additional settling motion of the heart observed by the abdominal retro-reflective marker. No image degradation compared to without the additional compensation was observed, strengthening our assumption that the discrepancy in rigid body motion observed in the anterior-inferior direction on the abdomen is directly related to internal organ motion in the head-to-feet direction. Although this motion present itself in a similar fashion as upward creep observed in exercise Tl-201, it is probably due to change in posture from upright to supine and not only due to a change in lung volume. Research Support: National Institute of Biomedical Imaging and Bioengineering (NIBIB), Grant No R01 EB001457. National Heart, Lung, and Blood Institute, Grant No R01 HL122484