Variational myocardial tracking from CINE-MRI with non-linear regularization

We present a new motion estimation approach for cardiac Magnetic Resonance Imaging (MRI) data from a variational framework. The improved performance of variational approach has been achieved by designing a new regularization term that properly handles motion discontinuities. This approach was applied to both synthetic and real data. The quantitative evaluation revealed the superior performance of the proposed method against reference approaches.