DERIVING PATIENT-SPECIFIC VENTRICULAR MECHANICS REFERENCE MODELS WITH APPLICATIONS IN THE ESTIMATION OF MYOCARDIAL STIFFNESS IN HEART FAILURE.
Zhinuo J. Wang1, Vicky Y. Wang1, Chris P. Bradley1, Alistair A. Young1,2, Jie J. Cao3, Martyn P. Nash1,2
1Auckland Bioengineering Institute, The University of Auckland, New Zealand; 2The University of Auckland, New Zealand; 3St. Francis Hospital, USA
Personalised computational models of the human left ventricle (LV) can provide estimates of myocardial tissue stiffness which are useful for investigating clinical hypotheses about the aetiology of heart failure. A significant limitation of such models has been the assumption that a stress-free mechanics reference geometry can be measured at some point during the cardiac cycle. This is typically measured at diastasis where the pressure inside the chamber is negligible. While this assumption may be reasonable for the healthy heart, it is problematic when dealing with heart failure (HF) patients since they often present with impaired relaxation kinetics and complications with hypertension, which means the LV never fully relaxes to its mechanical reference state during the cardiac cycle. We hypothesise that the assumption of negligible diastasis pressure masks important differences in model-based estimates of tissue mechanical properties between HF and control groups. This study presents a novel method for estimating the stress-free reference model of the LV and the passive tissue stiffness simultaneously. MRI and catheter pressure data recorded from HF patients were temporally aligned to provide simultaneous geometric and haemodynamic data at 3~6 time points from diastasis to end diastole. The reference LV model geometry was represented using principal component analysis dimension reduction techniques. The shape parameters of the reference model, and a tissue stiffness parameter were estimated simultaneously. The shape and stiffness parameters were optimised to minimise the projection between image-derived LV surface data and the model-predicted geometries at all time frames from diastasis to end diastole. The use of multiple time-points aims to improve the identifiability of the shape and stiffness parameters and alleviate their inter-dependence.
Tissue stiffness estimates were evaluated using the above approach on a per-patient basis, in order to investigate differences between different groups of heart failure patients and control subjects.