In this paper, we study the problem of pointwise motion tracking in echocardiographic images. We show the decorrelations between tissue motion and intensity variation is inevitable for certain kinds of tissue motion and decorrelation compensation is an ill-posed inverse problem if the decorrelation is beyond a certain correlation threshold. We compare the performance of different features using simulations and phantom examples. We find a threshold value of correlation coefficients below which the B-Mode signal works better than the radio frequency (RF) signal in the analysis of large deformation. We also demonstrate that the introduction of a quantitative reliability measure helps to improve the robustness of displacement estimation.