Cardiac PET/CT misregistration causes significant changes in estimated myocardial blood flow.

ABSTRACT

UNLABELLED Misregistration of cardiac PET/CT data can lead to misinterpretation of regional myocardial perfusion. However, the effect of misregistration on the quantification of myocardial blood flow (MBF) has not been studied. METHODS: Cardiac (82)Rb-PET/CT scans of 10 patients with normal regional myocardial perfusion were analyzed. Realignment was done for the baseline and stress PET/CT images as necessary, and MBF was obtained from dynamic data. Then, the stress images were misregistered by 5 mm along the x-axis (left) and z-axis (cranial) and again by 10 mm. A 10-mm misregistration in the opposite direction (-10 mm along the x-axis [right] and z-axis [caudal]) was also tested. Stress MBF was recalculated for 5-, 10-, and -10-mm misregistrations. RESULTS: Stress MBF of the left ventricle decreased by 10% ± 6% (P = 0.005) after 5-mm misregistration and by 24% ± 15% (P = 0.001) after 10-mm misregistration. In descending order, the most important stress MBF changes occurred in the anterior (39% ± 9%), lateral (34% ± 9%), apical (20% ± 16%), inferior (12% ± 10%), and septal (10% ± 12%) walls after 10-mm misregistration. Lesser changes were observed after 5-mm misregistration, with the same wall distribution. In contrast, -10-mm misregistration increased global MBF by 9% ± 6% (P = 0.004). In descending order, the overestimation of estimated MBF after -10-mm misregistration occurred in the lateral (15% ± 8%), apical (15% ± 18%), anterior (9% ± 5%), and inferior (9% ± 11%) walls. CONCLUSION: Misregistration of the stress PET/CT dataset leads to significant global and regional artifactual alterations in the estimated MBF. Quantitative error was observed throughout the myocardium and was not confined to those heart regions that extended into the lung on misregistered CT.