Differential Geometry provides a set of measures invariant under a set of transformations, in particular rigid, affine, and projective. The invariants by rigid motions are using almost all applications of computer graphics and geometric modeling. The affine case, since it is more general, allows to extend these tools. In this work, geometric properties are presented in the case of parametric or implicit surfaces, in particular the affine metric, the conormal and normal vectors, and the affine Gaussian and mean curvatures. Some usual results of Euclidean geometry, as the Minkowski formula, are extended for the affine case. This study allows to define estimators of affines structure in the case of isosurfaces. Although, the direct calculation of these structures greatly increases the number of operations and numerical instabilities. A geometrical reduction is proposed obtaining a much simpler and numerical stabler formulae. The geometrical properties are incorporated in the Marching Cubes algorithms, then they are analyzed and discussed.