Quantitative assessment of motion artifacts and validation of a new motion-correction program for myocardial perfusion SPECT

Patient motion during myocardial perfusion SPECT can produce images that sh ow artifactual perfusion defects. The relationship between the degree of mo tion and the extent of artifactual perfusion defects is not clear for eithe r single- or double-head detectors. Using both single- and double-head dete ctors and quantitative perfusion SPECT (QPS) software, we studied the patte rn and extent of defects induced by simulated motion and validated a new au tomatic motion-correction program for myocardial perfusion SPECT. Methods: Vertical motion was simulated by upward shifting of the raw projection data sets in a returning pattern (bounce) and in a nonreturning pattern at 3 dif ferent phases of the SPECT acquisition (early, middle, and late), whereas u pward creep was simulated by uniform shifting throughout the acquisition, L ateral motion was similarly simulated by left shifting of the raw projectio n datasets in a returning pattern and in a nonreturning pattern. Simulation s were performed using single- and double-head detectors, and simulated mot ion was applied to projection images from 8 patients who had normal Tc-99m- sestamibi SPECT findings. Additionally, images from 130 patients with actua l clinical motion were assessed before and after motion correction. The ext ent of perfusion defects was assessed by QPS, and a 20-segment, 5-point sco ring system was used to assess the effect of motion on the presence and ext ent of perfusion defects. Results: Of 12 bounce simulations, the bouncing m otion failed to produce significant (>3%) perfusion defects with either the single- or the double-head detector. With the single-head detector, early shifting created the largest defect, whereas with the double-head detector. shifting during the middle of the acquisition created the largest defect. With regard to upward creep, defects were of larger extent with the double- than the single-head detector. With the single-head detector, 8 of 20 simu lated motion patterns yielded significant perfusion defects of the left ven tricle, 7 (88%) of which were significantly improved after motion correctio n. With the double-head detector, 12 of 20 patterns yielded significant def ects, all of which improved significantly after correction. Of 2,600 segmen ts in the 130 patients with actual clinical motion, only 1.3% (30/2,259) of segments that were considered normal (score = 0 or 1) changed to abnormal (score = 2-4) after motion correction, whereas 27% (92/341) of abnormal seg ments were reclassified as normal after motion correction. Conclusion: Arti factual perfusion defects created by simulated motion are a function of the time, degree, and type of motion and the number of camera detectors. Appli cation of an automatic motion-correction algorithm effectively decreases mo tion artifacts on myocardial perfusion SPECT images.