Fast Predictive Multimodal Image Registration

TL;DR

This paper presents a deep learning approach for fast multimodal image registration that predicts deformation parameters, significantly reducing computation time while maintaining accuracy, and includes a probabilistic model for uncertainty estimation.

Contribution

The authors introduce a novel deep encoder-decoder network that predicts initial momentum for LDDMM registration, enabling rapid and accurate multimodal image registration with uncertainty quantification.

Findings

Achieves an order of magnitude faster registration than traditional methods.

Learned similarity measure improves registration consistency over mutual information.

Provides a Bayesian version for uncertainty estimation in registration.

Abstract

We introduce a deep encoder-decoder architecture for image deformation prediction from multimodal images. Specifically, we design an image-patch-based deep network that jointly (i) learns an image similarity measure and (ii) the relationship between image patches and deformation parameters. While our method can be applied to general image registration formulations, we focus on the Large Deformation Diffeomorphic Metric Mapping (LDDMM) registration model. By predicting the initial momentum of the shooting formulation of LDDMM, we preserve its mathematical properties and drastically reduce the computation time, compared to optimization-based approaches. Furthermore, we create a Bayesian probabilistic version of the network that allows evaluation of registration uncertainty via sampling of the network at test time. We evaluate our method on a 3D brain MRI dataset using both T1- and T2-weighted images. Our experiments show that our method generates accurate predictions and that learning the similarity measure leads to more consistent registrations than relying on generic multimodal image similarity measures, such as mutual information. Our approach is an order of magnitude faster than optimization-based LDDMM.