Quantifying the effect of vertical motion of patient during gated myocardial perfusion SPECT on indices of perfusion and function of left ventricle

Abstract

ObjectiveTo assess the effect of software-simulated “bouncing” motion on left ventricle (LV) perfusion and function indices concerning three main determinants of motion (duration, magnitude and time) by evaluating the sole effect and interaction of these attributes in a statistical model. MethodsTwenty-nine gated myocardial perfusion SPECT scans were selected for the study and then, “bounce” motion pattern was simulated manually regarding three main attributes of motion including duration (short versus long), magnitude (2 versus 4 pixels) and time (early versus late), all in upward vertical direction. All SPECT images are reconstructed and filtered with an identical method (OSEM algorithm) and same parameters. Indices of LV myocardial perfusion and function are derived using QGS package of Cedars-Sinai software in original and simulated-motion images and are then compared with each other. Two- and three-way Repeated Measure Within-Subjects ANOVA tests are conducted to evaluate the main effect of each variable or attribute and the interaction between them. ResultsSummed scores increase roughly exponentially from “no motion” to short bounce and then, to long bounce. In long 4-pixel bounce, perfusion defects are remarkable. All comparisons of defect extent (DE) and total perfusion deficit (TPD) are statistically significant. Mean difference between short bounce motion patterns with “no motion” is small even in 4-pixel movements (almost 3% or lower). In contrast, mean difference between long bounce motion patterns with “no motion” is higher than 5%. Using a paired-sample t-test, in all pairs, mean difference for ejection fraction (EF) is less than 4% which all are statistically significant. Value of end-diastolic volume (EDV) and end-systolic volume (ESV) are consistently decreased based on duration (from short to long) and magnitude (from 2 to 4 pixels). Using Within-Subjects ANOVAs, in long bounce, main effect of magnitude and interaction of magnitude and time, but not time solely, were statistically significant. In 2-pixel magnitude, none of variables and their interaction were significant, but in 4-pixel magnitude, EF showed statistical significance with duration. ConclusionThe perfusion parameters are to a higher extent involved by motion particularly in long bounce with a 4-pixel displacement. In short bounce, the effect is negligible, and therefore, no need to repeat the scan. Parameters of function are much less vulnerable to be affected by motion. Thus, contrary to current recommendations, there may be less need to repeat the scan in short 2-pixel bounce.