Inverse Consistency by Construction for Multistep Deep Registration
This work addresses the need for reliable and consistent image registration in medical imaging, though it appears incremental as it builds on existing neural registration methods with a specific structural constraint.
The authors tackled the problem of ensuring inverse consistency in neural image registration by proposing a simple technique that enforces this property by construction for networks parameterizing transforms via Lie groups, and extended it to multi-step registration for coarse-to-fine applications, achieving excellent registration accuracy on synthetic 2-D and four 3-D medical image tasks.
Inverse consistency is a desirable property for image registration. We propose a simple technique to make a neural registration network inverse consistent by construction, as a consequence of its structure, as long as it parameterizes its output transform by a Lie group. We extend this technique to multi-step neural registration by composing many such networks in a way that preserves inverse consistency. This multi-step approach also allows for inverse-consistent coarse to fine registration. We evaluate our technique on synthetic 2-D data and four 3-D medical image registration tasks and obtain excellent registration accuracy while assuring inverse consistency.