mlVIRNET: Multilevel Variational Image Registration Network
This addresses a specific bottleneck in medical image registration for lung analysis, though it appears incremental as it adapts conventional multilevel concepts to deep learning.
The paper tackles the limitation of existing deep learning registration methods to small deformations by introducing a multilevel framework that computes deformation fields at different scales, achieving significantly better results on inhale-to-exhale lung registration.
We present a novel multilevel approach for deep learning based image registration. Recently published deep learning based registration methods have shown promising results for a wide range of tasks. However, these algorithms are still limited to relatively small deformations. Our method addresses this shortcoming by introducing a multilevel framework, which computes deformation fields on different scales, similar to conventional methods. Thereby, a coarse-level alignment is obtained first, which is subsequently improved on finer levels. We demonstrate our method on the complex task of inhale-to-exhale lung registration. We show that the use of a deep learning multilevel approach leads to significantly better registration results.