As presented in previous editions, our method describes the three-dimensional reconstruction of the internal and external surfaces of the human's left ventricle from medical data. The reconstruction is a first process inside a complete VR application that will serve as an important diagnosis tool for hospitals. The segmentation process has been thoroughly refined in order to provide a robust border detection. Edges must be labeled as internal or external borders in order to allow the posterior dynamic method to perform the 3D reconstruction of the surfaces. The filtering process is based on the Maximum-Likelihood Classification approach. The algorithm rewards an edge depending on its probability to be a surface border. Probabilities are derived from several key characteristics.