A widely held goal in cardiac resynchronization therapy (CRT) is to implant the left-ventricular (LV) pacing lead in a late-activating region [1]. Speckle tracking echocardiography (STE) typically uses regional time to peak myocardial strain [2] rather than onset of strain to characterize regional dyssynchrony; however, electrical activation is directly related to the onset of contraction rather than the time to peak contraction. Preclinical MR strain imaging studies using myocardial tagging have demonstrated that strain MRI effectively detects the time of onset of circumferential shortening (TOS) [3]. The present study applied cine DENSE (Displacement Encoding with Stimulated Echoes) MR strain imaging [4] in CRT patients to test the hypothesis that the TOS correlates better with the electrical activation time compared with the time to peak circumferential shortening (TPS).Data were acquired from 50 patients with heart failure (HF) and left bundle branch block (LBBB) who met the standard criteria for CRT based on established clinical guidelines. Patients with and without myocardial scar were included. Cine DENSE imaging was performed on a 1.5T MRI system in three standard short-axis planes. Circumferential strain (E_cc) was computed using semiautomatic methods [5], and E_cc data were arranged into matrix form as a function of cardiac phase and LV segment. TPS was defined as the time point where E_cc reached its minimum value, and an active contour (AC) method was applied to the strain matrix to automatically detect the regional TOS, which was defined as the time point where dE_cc/dt first becomes negative (Fig. 1A, B). During the CRT implantation procedure, the electrical activation time from QRS onset to the electrogram at the LV lead implantation site (QLV) was measured. MRI and fluoroscopy registration methods were used to determine the LV lead position on the MR images [6]. The TOS and TPS were recorded for the LV segment corresponding to the LV lead position and then compared with the corresponding electrical activation time at the site of LV lead implantation.Fig. 1(B) shows an E_cc matrix from a mid-ventricular slice for a CRT patient with HF-LBBB and no scar. The figure illustrates the AC, which depicts regions of early and delayed onset of E_cc. Also evident is heterogeneity in the time to minimum E_cc. In the corresponding bull’s eye plots of TOS and TPS shown in Fig. 1(C, D), delayed TOS and TPS are observed in the lateral wall of the LV, and the CRT LV lead position is shown (black dot). Fig. 2 shows the correlations between electrical activation time and the mechanical parameters TOS (A, B) and TPS (C, D). In the datasets with no scar shown in Fig. 2(A, C), the R² was 0.78 (p<0.0001) for TOS versus electrical activation time, but only 0.32 (p = 0.004) for TPS versus electrical activation. In datasets with patients having the LV lead placed in a region containing scar shown in Fig. 2(B, D), the R² was 0.8 (p<0.0001) for TOS versus electrical activation time, but only 0.38 (p = 0.019) for TPS versus electrical activation time. We also observed that the slope of the regression line for TOS versus electrical activation is greater when the lead is placed in regions containing scar, as shown in Fig 2(B), versus without scar, as shown in Fig 2(A), consistent with scar-associated delays in electromechanical coupling.Regional mechanical activation times assessed using MRI cine DENSE TOS are strongly correlated with invasively measured electrical activation times, but this association was much weaker using MRI TPS. Cine DENSE imaging of TOS is a promising method for the noninvasive detection of late-activating segments in HF patients referred for CRT, which provides a distinct advantage for cine DENSE MRI over commonly used echocardiography methods.