The effects of extracellular contrast agent (Gadobutrol) on the precision and reproducibility of cardiovascular magnetic resonance feature tracking

Abstract

Background: Today feature tracking (FT) is considered to be a robust assessment tool in cardiovascular magnetic resonance (CMR) for strain assessment. The FT algorithm is dependent on a high contrast between blood pool and myocardium. Extracellular contrast agents decrease blood-myocardial contrast in SSFP images and thus might affect FT results. However, in a routine CMR scan, SSFP-cine images including short axis views are partly acquired after contrast agent injection. The aim of this study was to investigate the effect of extracellular contrast agent (Gadobutrol) (CA) on the precision and reproducibility of the feature tracking algorithm.

Methods: A total of 40 patient volunteers (mean age 51.2 ± 19 years; mean LVEF 61 ± 9 %) were scanned in supine position on a clinical 1.5 T MR scanner (Philips Ingenia). SSFP-cine images in midventricular short axis view (SA) as well as horizontal long axis view (HLA) were acquired before and 10-15 min after injection of a double dose Gadobutrol. FT derived systolic circumferential and longitudinal strain parameters were then calculated for pre- and post-contrast images.

Results: FT derived midventricular peak systolic circumferential strain (PSCS) (-24.8 ± 6.4 % vs. -20.4 ± 6.3 %), apical PSCS (-28.67 ± 6.5 % vs. -24.06 ± 8.5 %), basal PSCS (-24.42 % ± 6.5 vs. -20.68 ± 7.1 %), peak systolic longitudinal strain (-19.57 ± 3.3 % vs. -17.24 ± 4.1 %), midventricular epicardial PSCS (-9.84 ± 3.4 % vs. -8.13 ± 3.4 %) , midventricular PSCS-rate (-1.52 ± 0.4 vs. -1.28 ± 0.5) and peak diastolic circumferential strain rate (1.4 ± 0.5 vs. 1.05 ± 0.5) were significantly reduced after CA application. Post CA strain assessment showed higher intra- and interobserver variability. Pre-CA: intraobserver: mean 0.21, Limits of agreement (LoA) -2.8 and 3.2; interobserver: mean 0.64, LoA -2.8 and 4.1. Post-CA: intraobserver: mean -0.11, LoA -5.1 to 4.9; interobserver: mean 4.93 LoA 2.4 to 12.2.

Conclusion: The FT algorithm is dependent on a high contrast between blood and myocardium. Post CA strain results are significantly lower and less reproducible than pre-CA strain results.

Keywords: Cardiovascular magnetic resonance; Feature tracking; Gadobutrol; Myocardial strain; Robustness.

Fig. 1

Example of FT strain assessment in SSFP images pre-contrast (upper images) and post-contrast (lower images) in the same subject. The dotted line represents the propagated contour in a end-diastolic- and end-systolic phase. The coloured curves in the graph represent the segments of the midventricular slice

Fig. 2

Pearson correlation coefficient between pre-contrast and post-contrast derived strain (r = 0.81)

Fig. 3

Panel a Bland Altman plot of intraobserver variability for pre-contrast midventricular FT derived peak circumferential systolic strain (PSCS). Panel b Bland Altman plot of interobserver variability for pre-contrast midventricular FT PSCS. Panel c Bland Altman plot of intraobserver variability for post-contrast midventricular FT derived PSCS. Panel d Bland Altman plot of interobserver variability for post-contrast midventricular FT PSCS