Estimation of Left Ventricular Mechanical Activation Times from Motion-Corrected Cardiac 4DCT Images

Abstract

BackgroundMechanical activation times of the left ventricle (LV) are strongly correlated with patient response to cardiac resynchronization therapy (CRT); however, there remains an unmet clinical need for a simple, robust imaging modality that yields accurate and reproducible estimates of LV mechanical activation times. Modern four-dimensional computed tomography (4DCT) imaging systems with effective motion correction could fulfill this need. PurposeTo demonstrate the clinical utility of a newly developed cardiac motion correction algorithm called ResyncCT and to investigate its effect on estimating mechanical activation times of the LV. MethodsTwenty-four subjects with full cardiac volume, full cardiac cycle 4DCT images were retrospectively analyzed in this study. The reconstructed 4DCT images exhibited typical motion artifact characteristics that were dependent on the direction of LV wall motion with respect to the gantry position at each cardiac phase; these artifacts rotated synchronously with the gantry. The motion corrupted images, referred to as the uncorrected images, were processed with a novel cardiac motion correction algorithm called ResyncCT to yield motion corrected images (referred to as the ResyncCT images). Regional shortening (RSCT) of the LV was calculated over a 72-segment model for both the uncorrected and the ResyncCT images; each segment contained RSCT vs time data. From these RSCT vs time curves, LV mechanical activation was estimated as the time point at which the RSCT vs time curve shortened by 10% of its full dynamic range during systolic contraction; we refer to this time point as the time to onset of shortening (TOS). A shift in TOS > |35| ms has a significant effect on reclassifying patients based on their probabilities of responding to CRT; we investigated the effect of ResyncCT on the measured values of TOS over the LV. ResultsResyncCT had a pronounced effect on the TOS estimates; ResyncCT could potentially reclassify 23/24 (96%) subjects as either responders or non-responders to CRT. In 16/24 (67%) subjects, the differences in TOS that were sufficient to bring about a reclassification of CRT response were at least as large in surface area of the LV as one entire American Heart Association segment. ConclusionsWe demonstrated the clinical utility of ResyncCT in estimating LV mechanical activation times as times to onset of shortening in 24 human subjects. ResyncCT has a pronounced effect on the estimation of LV mechanical activation times; the differences in activation times between the ResyncCT and uncorrected images are heterogenous and subject specific. The effect of ResyncCT could potentially reclassify 96% of the subjects used in this study based on their probabilities of responding to CRT. The results reported in this study highlight the potential utility of ResyncCT in estimating timing of mechanical events of interest of the LV for CRT planning.