Various medical treatments require an accurate determination of the shape of a considered anatomic structure. The shape is often recovered from several delineations performed on a two-dimensional gold standard imagery. Using true 3D imagery is attractive to supplement this gold standard. However, before using 3D modalities in clinical routine, it must be proved that these modalities are well suited to the delineation task. We propose in this paper a methodology for validating a new imaging modality with respect to a reference imagery.