Symmetrization of the Non-rigid Registration Problem Using Inversion-Invariant Energies: Application to Multiple Sclerosis

摘要: Without any prior knowledge, the non-rigid registration of two images is a symmetric problem, i.e. we expect to find inverse results if exchange these images. This symmetry nonetheless broken in most intensity-based algorithms. In this paper, explain reasons why algorithms are asymmetric. We show that asymmetry quadratic regularization energies causes an oversmoothing expending regions relatively shrinking regions, hampering particular registration-based detection evolving processes. therefore propose use inversion-invariant energy symmetrize problem. To minimize energy, methods used, depending on whether compute transformation or not. Finally, illustrate interest theory using both synthetic and real data, improve segmentation lesions MR patients suffering from multiple sclerosis.