Multi-scale Neural ODEs for 3D Medical Image Registration

TL;DR

This paper introduces a multi-scale neural ODE framework for 3D medical image registration that combines the accuracy of iterative optimization with the speed of deep learning, and includes a modality-independent similarity metric.

Contribution

It proposes a novel neural ODE-based registration optimizer that is faster and more adaptable, along with a modality-independent similarity metric for multi-contrast images.

Findings

Outperforms existing methods in accuracy and speed

Effective across different image contrasts

Validated on public and private datasets

Abstract

Image registration plays an important role in medical image analysis. Conventional optimization based methods provide an accurate estimation due to the iterative process at the cost of expensive computation. Deep learning methods such as learn-to-map are much faster but either iterative or coarse-to-fine approach is required to improve accuracy for handling large motions. In this work, we proposed to learn a registration optimizer via a multi-scale neural ODE model. The inference consists of iterative gradient updates similar to a conventional gradient descent optimizer but in a much faster way, because the neural ODE learns from the training data to adapt the gradient efficiently at each iteration. Furthermore, we proposed to learn a modal-independent similarity metric to address image appearance variations across different image contrasts. We performed evaluations through extensive experiments in the context of multi-contrast 3D MR images from both public and private data sources and demonstrate the superior performance of our proposed methods.