Unsupervised segmentation and personalised FE modelling of in vivo human myocardial mechanics based on an MRI atlas

We have developed techniques to automatically generate personalised biomechanical models of patients´ hearts based on 3D cardiac images. We demonstrate this approach using multi-slice computed tomography images. Unsupervised segmentation was performed using non-rigid image registration with a segmented image. A finite element model was automatically fitted to the segmented data of the left ventricle. Passive and contractile myocardial mechanical properties were tuned to match the segmented surface geometries at end-diastole and end-systole, respectively. Global and regional indices of myocardial mechanics, including cardiac wall and myofibre strain distributions, were then quantified. This automated biomechanical modelling approach to cardiac image analysis provided noninvasive methods to characterise heart function, and may provide new quantitative diagnostic markers for heart failure.