Accurate and Efficient 3D Reconstruction of Right Heart Shape and Motion from Multi-Series Cine-MRI

Abstract The accurate reconstruction of the right heart geometry and motion from time-resolved medical images enhances diagnostic tools based on image visualization as well as the analysis of cardiac blood dynamics through computational methods. Due to the peculiarity of the right heart morphology and motion, commonly used segmentation and/or reconstruction techniques, which only employ Short-Axis cine-MRI, lack accuracy in relevant regions of the right heart, like the ventricular base and the outflow tract. Moreover, the reconstruction procedure is time-consuming and, in the case of the generation of computational domains, requires a lot of manual intervention. This paper presents a new method for the accurate and efficient reconstruction of the right heart geometry and motion from time-resolved MRI. In particular, the proposed method makes use of surface morphing to merge information coming from multi-series cine-MRI (such as Short/Long-Axis and 2/3/4 Chambers acquisitions) and to reconstruct important cardiac features. It also automatically provides the complete cardiac contraction and relaxation motion by exploiting a suitable image registration technique. The method is applied both to a healthy and a pathological (tetralogy of Fallot) case, and yelds more accurate results than standard procedures. The proposed method is also employed to provide significant input for computational fluid dynamics. The corresponding numerical results demonstrate the reliability of our approach in the computation of clinically relevant blood dynamics quantities.