Variational implicit surfaces have been widely used in computer graphics and animation and generally provide solutions for uniformly distributed sparse data. The fields of medicine and biology increasingly rely on accurate three-dimensional reconstructions. Despite the fact that biological structures fit perfectly within the scope of variational implicit interpolation (smooth closed surfaces), the data to model in these fields constitute a computational challenge (sparse non-homogeneous data, high curvatures). Several parametric and simplicial approaches have been presented[1, 2]. This paper presents an approach for creating and optimizing 3D reconstructions based on variational implicit surfaces for a dataset consisting of serial parallel sections. Preliminary results and perspectives are presented.