Echography (ultrasound imaging of the heart) is one of the driving application areas of medical imaging. Our focus is to track the motion of endocardium during a complete cardiac cycle to allow clinicians for estimating the left ventricle and atrium (LVA) area deformations. The main challenge is to do this in a suitable response time. We describe a PETSc-based parallel software designed to detect and delineate the LV contour during a complete cardiac cycle. LV segmentation is performed by applying a three steps algorithm: speckle reduction, optic flow computation and spatio-temporal level-set segmentation. Using a scale-space approach each step is mathematically described by nonlinear time dependent PDE. Experiments on real data are obtained on a parallel computing platform made of 16 blades Dell PowerEdge M6000 each one consisting of 2 processors quad-core (Intel Xeon E5410@2.33GHz (64 bit)) connected by high performance network InfiniBand and located at the University of Naples Federico II.