Author + information
- Brett R. Cowan, BE, MB, ChB,
- Sophie M. Peereboom, BSc,
- Andreas Greiser, PhD,
- Jens Guehring, PhD and
- Alistair A. Young, PhD∗ ()
- ↵∗Department of Anatomy with Radiology, Faculty of Medical and Health Sciences, University of Auckland, 85 Park Road, Auckland 1142, New Zealand
Myocardial circumferential strain is a direct measure of tissue function that, unlike ejection fraction, is not directly influenced by ventricular size and shape. Strain estimated from standard cine cardiac magnetic resonance (CMR) steady-state free precession (SSFP) images using image feature tracking has excellent global agreement with CMR tagging over the whole slice (1). However, segmental strains show large differences (1). We investigated the determinants of segmental SSFP strain in 85 patients (age 67 ± 6 years, 63 men) who participated in the ONTARGET (ONgoing Telmisartan Alone and in Combination With Ramipril Global Endpoint Trial) study (2). Written informed consent was obtained, and the institutional review board approved the study. Retrospectively gated SSFP cine images and prospectively gated tagged gradient echo (FLASH) images were acquired in the same short-axis locations (n = 414 slices in total) on a 1.5-T Siemens Avanto scanner (Siemens Healthcare, Erlangen, Germany).
Epicardial and endocardial contours were manually placed on the first (end-diastolic) frame of each short-axis slice (Figure 1A). A generalized feature tracking and strain analysis framework (3) was used to track motion and calculate strain in both SSFP and tagged images. Epicardial and endocardial contours were obtained for each frame by tracking the end-diastolic contours. The SSFP images were then modified by: 1) replacing all pixels between the epicardial and endocardial contours with a constant value (Myo); 2) replacing all pixels outside the myocardium with a constant value (ExMyo); and 3) replacing both regions with constant values, leaving contour information only (Contour) (Figure 1A). Fiducial landmarks placed at the anterior and posterior insertions of the right ventricular (RV) free wall into the septum were used to define 6 segments per slice.
Global circumferential strain showed very good agreement between tagged and untagged estimates, even if all image features except the contours were masked (Figure 1B). However, regional bias in segmental strain was found, with the septal wall underestimated and the lateral wall overestimated compared with tagging (analysis of variance [ANOVA] p < 2 × 10−16). These differences were larger when the image outside the myocardium was masked (ANOVA p < 0.02). The agreement between original and masked segmental SSFP segmental strain was worse when the image outside the myocardium was masked, whereas masking the myocardium made little difference (Figure 1C).
The variability (standard deviation of the differences) between tagging and SSFP strain was smallest for the global estimate, and increased for the segmental estimates for all SSFP estimates (5% vs. 9%, respectively). The variability between masked and original SSFP estimates was largest for ExMyo and Contour estimates.
Through-plane motion is likely to introduce errors in strain. Because tags are created orthogonal to the image plane, the tagging strain estimate is more robust to through-plane motion. A 3-dimensional analysis may give better agreement between untagged and tagged estimates. Limitations include differences in slice thickness (6 mm for SSFP and 8 mm for tagging), which may give rise to small differences in strain. Masking the myocardium also introduces different endocardial and epicardial image edge features, which may influence the feature tracking results. However, Figure 1B shows that the Myo mask did not significantly change segmental estimates.
Our results support the conclusions that global strain is primarily determined by changes in the circumference, and that features outside the myocardium greatly influence segmental strain estimates. Because the anterior insertion point of the RV free wall into the interventricular septum curves around the left ventricle, through-plane motion gives rise to apparent deformation. Also, relative motions of papillaries and trabeculae give rise to strain artifacts. The pericardium also creates a potential discontinuity between myocardial and extracardiac motion.
Please note: This work was funded by Siemens AG and The Health Research Council of New Zealand. Drs. Cowan and Young have received consulting fees from Siemens AG. Drs. Greiser and Guehring are employees of Siemens AG. Ms. Peereboom was supported by an internship from the Netherlands Heart Foundation and Eindhoven University of Technology.
- 2015 American College of Cardiology Foundation
- Cowan B.R.,
- Young A.A.,
- Anderson C.,
- et al.