Efficient multiscale deformable registration frameworks are proposed by combining edge preserving scale space (EPSS) with the free form deformation (FFD) for registration of medical images, where multiscale edge information can be used for optimizing the registration process. EPSS which is derived from the total variation model with the L1 norm (TV-L1) can provide useful spatial edge information for mutual information (MI) based registration. At each scale in registration process, the selected edges and contours are sufficiently strong to drive the deformation using the FFD grid, and then the deformation fields can be gained by a coarse to fine manner. In our deformable registration framework, two ways are proposed for implementing this idea. The experiments on clinical images including PETCT and CT-CBCT show accuracy and robustness when compared to traditional method for medical imaging system.