This work takes is part of a medical research project which intends to induce and study cardiac hibernation in rats. The underlying goal is to understand the physiology of heart disease. We present here a novel method to compute the 2D-deformation field of the heart (rat or human) from tagged MRI. Previous work is not suitable for wide clinical use for different reasons, including important computing time and lack of robustness. We propose an original description of tags as local minima of 1D signals. This leads us to a new formulation of the tag tracking problem as an Extrema Temporal Chaining (ETC) and a 2D-rendering. 2D-displacements are then interpolated on a dense field. The developed method is fast and robust. Its performances are compared to those of HARP, a leading method in this field.