Improving Understanding of Long-Term Cardiac Functional Remodelling via Cross-Sectional Analysis of Polyaffine Motion Parameters

Abstract

Changes in cardiac motion dynamics occur as a direct result of alterations in structure, hemodynamics, and electrical activation. Abnormal ventricular motion compromises long-term sustainability of heart function. While motion abnormalities are reasonably well documented and have been identified for many conditions, the remodelling process that occurs as a condition progresses is not well understood. Thanks to the recent development of a method to quantify full ventricular motion (as opposed to 1D abstractions of the motion) with few comparable parameters, population-based statistical analysis is possible. A method for describing functional remodelling is proposed by performing statistical cross-sectional analysis of spatio-temporally aligned subject-specific polyaffine motion parameters. The proposed method is applied to pathological and control datasets to compare functional remodelling occurring as a process of disease as opposed to a process of ageing.

Cardiac Motion Evolution Model for Analysis of Functional Changes Using Tensor Decomposition and Cross-Sectional Data

“…In terms of describing the full functional dynamics, the problem requires modelling 4D objects; 3D in shape + time over the cardiac cycle (ms), and projecting this over long-term time intervals (years). As an extension of this to cardiac function analysis, cross-sectional analysis of matrices describing cardiac motion was recently proposed in [23]. Methods used previously for statistical analysis of cardiac images are described in the next section.…”

“…To our knowledge there has been no use of higher-order PLS applied to cardiac image analysis. We extend on the work of [23], in which matrices were regressed using standard PLS, to perform higher-order PLS on the full (rather than matricised) tensors containing the motion parameters.…”

“…The relationship is inversed using canonical correlation analysis (CCA) to obtain a generative (static) shape evolution model. This method (PCA+CCA) was recently extended to the application of cardiac motion analysis applied to matricised tensors in [23].…”