Two algorithms for image reconstruction in electromagnetic imaging are proposed. The first approach concerns the application of a hybrid version of the genetic algorithm to tomographic imaging of dielectric configurations. In the second approach, buried inhomogeneities are schematized as multilayer infinite dielectric cylinders with elliptic cross sections. An efficient recursive analytical procedure is used for the forward scattering computation. A cost function is constructed in which the field is expressed in series solution of Mathieu functions. Starting by the input scattered data, the iterative minimization of the functional is performed by a new optimization method called memetic algorithm.