Quick Fire Session
SCMR 22nd Annual Scientific Sessions
Individualized hemodynamic models of the cardiovascular system can give improved insight into the interactions between different parts of the cardiovascular system, enable computation of hemodynamic variables that are difficult or impossible to measure experimentally, and allow for prediction of intervention outcomes. These models are typically individualized using a series of inputs obtained from experimental measurements, such as cardiac magnetic resonance (CMR) acquisitions. Assessing how the variability of these input data influences the subject-specific parameter estimations is critical for obtaining reliable biomarkers and successfully applying them in the clinical setting (2). In this study, we evaluated the reproducibility of model-based parameters describing left ventricular time-varying elastance and aortic compliance obtained using a previously developed approach (3).
The approach combines flow and morphological data from CMR acquisitions and brachial cuff pressure measurements with a lumped-parameter model of the cardiovascular system. The reproducibility of the parameters with respect to intra-, inter-observer and inter-sequence variability in the input measurements was assessed in a group of ten healthy subjects. For this purpose, the subjects underwent CMR examinations to acquire 4D Flow CMR data and 2D cine morphological images, subsequently analyzed by two different observers. To analyze inter-sequence variability, the 4D Flow CMR data were acquired using two different sequences, a spoiled gradient echo (SGRE) based sequence and an echo-planar imaging (EPI) based sequence.
Among the studied parameters, the diastolic time constant of the left ventricle had the lowest coefficient of variation in the intra- and inter-observer analysis (2.5 and 3.9%, respectively) (Table 1), while the highest coefficient of variation was found for the compliance of the ascending aorta (21.8 and 34.9%, respectively). The rest of the parameters showed good to moderate variability in the intra-and inter-observer studies, with coefficients of variation in the range between 4.7 and 19.5%. In comparing parameters estimated using data from the two 4D Flow CMR sequences, the coefficients of variation ranged between 3.6 and 41%, being the diastolic time constant of the left ventricle and the compliance of the ascending aorta the parameters with the lowest and the highest variability, respectively (Table 2).
The modelling approach allows for estimating left ventricular elastance parameters and aortic compliance from non-invasive measurements with good to moderate reproducibility with respect to intra-, inter-user and inter-sequence variability.