ABSTRACT
Cardiovascular disease is still the leading cause of mortality in the world. In recent years, there has been an explosion of cardiovascular imaging data available to researchers, enabling “big data” techniques to characterize cardiovascular disease at the population level, rather than the individual patient. In this chapter, we show techniques that allow for the reduction of millions of pixels down to a few hundred cardiac shape parameters, for each patient. We discuss how to exploit large magnetic resonance image databases by finite-element modelling and statistical shape analysis. These methods can be applied to evaluate and quantify progression of disease. Most applications have typically focused on the left ventricle since this is most affected in coronary artery disease. However we also discuss the extensions of these methods to modelling both left and right ventricles in congenital heart disease, the most common type of birth defect. We show methods that significantly decrease analysis time while increasing the amount of data accumulated. We envisage that these methods will form the basis of future computer-aided diagnostic tools.
