Objective: The progress in computed tomography (CT) has improved temporal resolution and shortened the acquisition time. We compared cardiac function using 64-slice CT with left ventriculography (LVG) and cardiovascular magnetic resonance (CMR).
Methods: A head-to-head comparison between CT, LVG and CMR was performed in 41 patients. In global LV function, CMR served as the reference. Regional wall motion was compared in a 5-point scoring system.
Results: CT had excellent intra- and interobserver reproducibility. Ejection fraction, end-diastolic and end-systolic volumes by CT were closely correlated with CMR (r = 0.95, 0.96 and 0.98, respectively), while LVG underestimated LV volumes (p < 0.01). The standard deviation of ejection fraction difference between CT and CMR was significantly lower than that between LVG and CMR (p = 0.0015). In regional function, there were good agreements of 94.8% (kappa = 0.82) between CT and LVG and 94.5% (kappa = 0.84) between CT and CMR. The intermethod agreements in mild hypokinesis using CT tended to be lower.
Conclusion: An excellent correlation was observed between CT and CMR in the LV function over a wide range of heart rates. However, even though 64-slice CT tended to be less sensitive in detecting mild hypokinesis, it still showed excellent concordance in advanced regional abnormalities.