A novel imaging strategy for structural segmentation of the left ventricle

The combination of computational fluid dynamics and in vivo imaging techniques such as cardiovascular magnetic resonance (CMR) allows the calculation of blood flow indices that cannot be measured directly. This combined strategy requires the provision of accurate anatomical delineation of cardiovascular structure as well as boundary flow conditions. Thus far, an automatic method for structure segmentation has yet to be developed. The purpose of this paper is to present a novel imaging strategy, based on contrast enhancement characteristics of TrueFISP CMR, that allows the signal intensities of blood and myocardium of the Left Ventricle (LV) to be extracted without user interaction. This has been used to provide a fully automatic segmentation of the LV endocardial border allowing the creation of models suitable for CFD analysis. The method has been validated with data acquired from 10 asymptomatic subjects. The results have shown to be comparable to that of manual delineation by experienced observers.