FOD peaks do not have ‘eigenvalues’. Eigenvalues can be calculated from the diffusion tensor model due to the nature of that model as a tensor matrix. The only possible ‘analogues’ in the context of the spherical deconvolution model would be the FOD peak amplitude / fixel integral; but it is quite a long stretch to make that comparison. They also do not intrinsically lend themselves to the generation of a “map like FA”, since these provide a measure per fixel rather than per voxel. There have been multiple proposals for a “generalised FA” measure that quantify the anisotropy of the fibre configuration in each voxel without explicitly invoking the diffusion tensor model, but these all have the drawback that they are not fibre-specific.
It might be worth reading something like Flavio’s recent SD review paper to better understand the underlying concepts.