ASymReg: Robust symmetric image registration using anti-symmetric formulation and deformation inversion layers

Deep learning based deformable medical image registration methods have emerged as a strong alternative for classical iterative registration methods. However, the currently published deep learning methods do not fulfill as strict symmetry properties with respect to the inputs as some classical registration methods, for which the registration outcome is the same regardless of the order of the inputs. While some deep learning methods label themselves as symmetric, they are either symmetric only a priori, which does not guarantee symmetry for any given input pair, or they do not generate accurate explicit inverses. In this work, we propose a novel registration architecture which by construction makes the registration network anti-symmetric with respect to its inputs. We demonstrate on two datasets that the proposed method achieves state-of-the-art results in terms of registration accuracy and that the generated deformations have accurate explicit inverses.

Deformation equivariant cross-modality image synthesis with paired non-aligned training data

Cross-modality image synthesis is an active research topic with multiple medical clinically relevant applications. Recently, methods allowing training with paired but misaligned data have started to emerge. However, no robust and well-performing methods applicable to a wide range of real world data sets exist. In this work, we propose a generic solution to the problem of cross-modality image synthesis with paired but non-aligned data by introducing new deformation equivariance encouraging loss functions. The method consists of joint training of an image synthesis network together with separate registration networks and allows adversarial training conditioned on the input even with misaligned data. The work lowers the bar for new clinical applications by allowing effortless training of cross-modality image synthesis networks for more difficult data sets and opens up opportunities for the development of new generic learning based cross-modality registration algorithms.