Myocardial delayed enhancement with dual-source CT: Advantages of targeted spatial frequency filtration and image averaging over half-scan reconstruction

Abstract

BackgroundClinical utility of myocardial delayed-enhancement CT is currently limited because of relatively poor contrast-to-noise ratio (CNR) and artifacts. Targeted spatial-frequency filtration (TSFF) is a hybrid algorithm of half- and full-scan reconstruction that can achieve both high temporal resolution and improved stability of myocardial signal. ObjectiveThe purpose of this study was to evaluate image quality of delayed-enhancement CT using TSFF with image averaging and its reproducibility in infarct assessment in comparison with conventional half-scan reconstruction (HALF). MethodsForty patients with suspected coronary artery disease underwent delayed-enhancement CT with HALF and TSFF using dual-source CT. Two blinded readers independently determined the presence and size of delayed enhancement. Image quality, signal-to-noise ratio and CNR were assessed. The presence of delayed enhancement on CT was compared with magnetic resonance imaging in 12 patients. ResultsTSFF with averaging of 4 image stacks acquired during 1 breathhold demonstrated significantly better image quality compared with HALF. Good left ventricular lumen-myocardium contrast was consistently achieved with TSFF in patients who received iodine dose of >600mgI/kg. The signal-to-noise ratio and CNR were 11.3 ± 4.2 and 4.5 ± 1.6 by TSFF, being significantly higher than those by HALF (7.9 ± 2.9 and 3.3 ± 1.8; P < .01 for both). Interobserver reproducibility of infarct sizing was markedly improved by using TSFF instead of HALF (intraclass correlation coefficient: 0.86 vs 0.50). Agreement with magnetic resonance imaging by kappa statistics was 0.85 with TSFF and 0.74 with HALF. ConclusionsTSFF with image averaging can significantly improve image quality of delayed-enhancement CT and considerably enhances interobserver reproducibility of infarct sizing.