Effects of time discrepancies between input and myocardial time-activity curves on estimates of regional myocardial perfusion with PET

Abstract

Estimates of myocardial perfusion with PET using kinetic models require faithful recording of radioactivity content in blood and myocardium. Typically the arterial time-activity curve is obtained by placing a region of interest (ROIs) within the left atrial or left ventricular cavity. However, curves generated from these regions appear earlier in time than tissue time-activity curves obtained from ROIs within the myocardial tissue, and such time discrepancies can lead to errors in flow estimates.

Methods: The magnitude of these time discrepancies and their effect on estimates of regional myocardial perfusion using oxygen-15-water were measured in 30 normal subjects evaluated at rest and again after administration of dipyridamole.

Results: Under baseline conditions, the left atrial curve appeared 0.97 +/- 0.67 (s.d.) before the ascending aorta input curve (p < 0.05) and estimated perfusion decreased from 1.28 +/- 0.28 ml/g/min using the left atrial curve uncorrected for time to 0.98 +/- 0.27 ml/g/min after correction (p < 0.05). After dipyridamole, the left atrial curve appeared 0.68 +/- 0.72 sec before the ascending aorta curve (p < 0.05) and estimated perfusion decreased from 3.60 +/- 1.40 ml/g/min using the left atrial curve uncorrected for time to 3.24 +/- 1.26 ml/g/min using the time-corrected curve (p < 0.05). Because the magnitude of time discrepancies between the left ventricular and ascending aortic curves was less (0.25 +/- 0.34 and 0.19 +/- 0.23 sec at rest and after dipyridamole, respectively), effects on flow estimates were more modest.

Conclusions: The results of this study demonstrate that time discrepancies between input and tissue time-activity curves can affect estimates of myocardial flow. Correction for this potential source of error is proposed.