Respiratory Motion Alters Measurement of SPECT Myocardial Blood Flow

Background: Respiratory motion degrades image quality in nuclear medicine imaging. Motion during acquisition blurs the image, increasing partial volume effects, and can increase the spill-over contamination from hot extra-cardiac structures into the myocardial wall. Studies of static myocardial perfusion images have shown that there is substantial motion (>1cm) in up to 33% of patient studies. It is thus expected that there is respiratory motion within dynamic SPECT myocardial blood flow (MBF) studies. In addition, PET studies have shown that patient motion can have a significant impact on MBF measurements. The goal of our study was to estimate the prevalence and magnitude of respiratory motion in SPECT MBF studies and determine the impact on measured MBF values.

Methods: Data from 25 patients were retrospectively analyzed. SPECT MBF was measured using a one-day rest/stress protocol with Tc99m-tetrofosmin on a pinhole cardiac SPECT camera. The imaging was repeated for each patient at an interval of 17 ± 13 days. Each of the 100 dynamic scans was divided into 19 frames of 9 x 10 sec, 6 x 15sec, 4 x 120sec. For each dynamic frame, the acquired count rate in each 100 msec interval was determined from the listmode data and data were sorted into 10 separate respiratory gates based on this count rate. The gates were reconstructed independently and the images registered using rigid-body translation to determine the motion vector between gates. The original projection data were then reconstructed offline with a 4D iterative algorithm that incorporated the motion vectors to generate a single respiratory-motion-corrected (RMC) image for each dynamic frame. A 3D version of the same algorithm was used to reconstruct images without motion correction (NMC). The maximum distance between any two gates, averaged over the last 4 frames (between 3 and 11 min after tracer injection), was used as a measure of patient motion. Three patients with large motion (≥1cm, mean 1.5cm) at both rest and stress, and 3 patients with minimal motion (≤ 3mm), were processed to determine MBF using 4DM reserve (INVIA) for both NMC and RMC reconstructions.

Results: The average patient motion measured in the dynamic studies was 6.7mm ± 4.0 mm (maximum 25mm). Of the 100 scans, 17 had motion >10mm. The patient motion in the 1^st study was not significantly different from that in the 2^nd study (6.9mm vs 6.4mm, p=0.14) but was correlated between the 2 studies (r=0.86). The difference in motion between the two studies averaged -0.5mm ± 2.4mm (max difference = 9mm). Motion at stress was slightly, but significantly, larger than at rest (7.1mm vs 6.3mm, p=0.04) and correlated between rest and stress (r=0.76). For the 3 low-motion rest-stress studies, the average difference in global flow between NMC and RMC reconstructions was 10%, 14% and 8.5% for rest MBF, stress MBF, and myocardial flow reserve (MFR) respectively. The corresponding average regional values in the LAD, LCx, and RCA territories were 11%, 16%, and 11%. These differences are consistent with previously measured intra-user variation of ~15%. For the 3 high-motion rest-stress studies, the average difference in global flow between NMC and RMC reconstructions was 37%, 21% and 17% for rest MBF, stress MBF, and myocardial flow reserve (MFR) respectively. The corresponding average regional values in the LAD, LCx, and RCA territories were 39%, 23%, and 28%. Conclusions: Respiratory motion caused displacement of the heart by >10mm in 17% of dynamic scans. Respiratory motion can substantially alter SPECT MBF and MFR values.