Left ventricular strain and its pattern estimated from cine CMR and validation with DENSE

Abstract

Measurement of local strain provides insight into the biomechanical significance of viable myocardium. We attempted to estimate myocardial strain from cine cardiovascular magnetic resonance (CMR) images by using a b-spline deformable image registration method. Three healthy volunteers and 41 patients with either recent or chronic myocardial infarction (MI) were studied at 1.5 Tesla with both cine and DENSE CMR. Regional circumferential and radial left ventricular strains were estimated from cine and DENSE acquisitions. In all healthy volunteers, there was no difference for peak circumferential strain (- 0.18 ± 0.04 versus - 0.18 ± 0.03, p = 0.76) between cine and DENSE CMR, however peak radial strain was overestimated from cine (0.84 ± 0.37 versus 0.49 ± 0.2, p < 0.01). In the patient study, the peak strain patterns predicted by cine were similar to the patterns from DENSE, including the strain evolution related to recovery time and strain patterns related to MI scar extent. Furthermore, cine-derived strain disclosed different strain patterns in MI and non-MI regions, and regions with transmural and non-transmural MI as DENSE. Although there were large variations with radial strain measurements from cine CMR images, useful circumferential strain information can be obtained from routine clinical CMR imaging. Cine strain analysis has potential to improve the diagnostic yield from routine CMR imaging in clinical practice.

A Novel Method for Estimating Myocardial Strain: Assessment of Deformation Tracking Against Reference Magnetic Resonance Methods in Healthy Volunteers

“…have developed a method of estimating myocardial strain designed to overcome some of the theoretical limitations of feature-tracking, which uses local features, by estimating pixel-wise strain for myocardial deformation incorporating both local and global myocardial features from cine images. Deformation-tracking21 is based on an in-house intensity-based b-spline deformable registration method22. Our prelimary study21 and other studies2324 have demonstrated that deformable image registration methods could be an alternative way to quantify regional LV deformation.…”

“…Deformation-tracking21 is based on an in-house intensity-based b-spline deformable registration method22. Our prelimary study21 and other studies2324 have demonstrated that deformable image registration methods could be an alternative way to quantify regional LV deformation.…”

“…in Matlab (Mathworks, Natick, US), in line with a standard, automated protocol that minimised variability and bias (H.G. )21. In the deformation-tracking approach21 (Fig.…”

“…)21. In the deformation-tracking approach21 (Fig. 2), an intensity-based b-spline deformable registration method22 was implemented to deform one target image into a reference image by minimizing the sum of squared differences of the two images, the deformation fields were represented by cubic b-spline basis functions, and further smoothed by introducing a regularized term in the minimization procedure.…”

Passive diastolic modelling of human ventricles: Effects of base movement and geometrical heterogeneity

“…In addition, due to lack of 268 patterns/features in cine images, higher uncertainties presents while estimating pixel-wise 269 strain. Gao et al (2014) showed that the regional circumferential strains could be estimated 270 correctly from cine images. However, greater discrepancies exists during the estimation of 271 regional radial strains, and therefore, could not be used in FE modelling (Gao et al, 2015).…”