SCMR 22nd Annual Scientific Sessions
Myocardial strain is an important parameter for the evaluation of myocardial regional function and early subtle dysfunction. As anatomically known, the left ventricular (LV) myocardial wall comprises endo, mid and epicardial three layers and different etiologies and pathogenesis could affect the three myocardial layers to different extent1. At present, speckle tracking of echocardiography (STE) is the main approach for layer-specific myocardial deformation analysis234and has demonstrated the unique values of layer-specific measurement536. Despite CMR has the advantage of higher spatial resolution and better reproducibility, the utilization of layer-specific deformation based on CMR are limited. DRA (deformation registration algorithm) was a potential technique to analyze layer specific strain based on balanced stead state free precession (bSSFP) cine. Validation of DRA with current gold standard technique such as displacement encoding with stimulated echoes (DENSE) was required.
Twenty healthy volunteers and 30 cardiomyopathy patients underwent 3.0T cardiac magnetic resonance imaging (CMR). The bSSFP cine and DENSE cine images were obtained on three position-matched short-axis slices (basal, mid-ventricle and apical levels). Left ventricular global and layer-specific circumferential strain (Ecc) were quantified by DRA (Trufistrain 2.0, Siemens Healthcare, Princeton, NJ) andDENSEanalysismethods (https://github.com/denseanalysis/denseanalysis). Global Ecc measurements regarded myocardial wall as an integral, while the layer-specific Eccmeasured the myocardial deformation trisectly and separately. The transmural gradient from endowall to epiwall (Gendo-epi) was calculated as the absolute difference between the Ecc value on endo and epi layers. The correlation between Ecc values derived by two methods was evaluated by spearman correlation coefficient (r) and intra-class correlation coefficient (ICC). The intra- and inter observer reproducibility of two methods was evaluated bycoefficients of variations (CoVs).
For global Ecc, apical level had relatively lower agreement between DRA and DENSE compared with basal and mid-ventricle levels (r = 0.874 [apical] vs. r = 0.921 [basal] and r = 0.940 [mid-ventricle], all p < 0.001). For layer-specific Ecc, the endo layer showed slightly lower agreement compared with mid and epi-layer (r = 0.880 [endo] vs. r = 0.908 [mid] and r = 0.891 [epi], p < 0.001). The transmural gradient from endowall to epiwall on three levels also showed good correlation by two methods (r = 0.595 [basal], r = 0.751 [mid-ventricle], r = 0.708 [apical], p<0.001). Both DRA and DENSE methods represented good intra- and inter observer reproducibility.
DRA is a feasible and robust technique for global and layer-specific Ecc analysis validated with DENSE.