Embedding Gradient-based Optimization in Image Registration Networks

TL;DR

This paper introduces GraDIRN, a deep learning model that incorporates traditional gradient-based energy optimization into its architecture, achieving superior image registration performance with fewer parameters and enhanced robustness.

Contribution

It presents a novel neural network that unrolls multiresolution gradient-based optimization, bridging traditional methods and deep learning for image registration.

Findings

Achieved state-of-the-art registration accuracy.

Used fewer parameters than existing methods.

Demonstrated good data efficiency and robustness.

Abstract

Deep learning (DL) image registration methods amortize the costly pair-wise iterative optimization by training deep neural networks to predict the optimal transformation in one fast forward-pass. In this work, we bridge the gap between traditional iterative energy optimization-based registration and network-based registration, and propose Gradient Descent Network for Image Registration (GraDIRN). Our proposed approach trains a DL network that embeds unrolled multiresolution gradient-based energy optimization in its forward pass, which explicitly enforces image dissimilarity minimization in its update steps. Extensive evaluations were performed on registration tasks using 2D cardiac MR and 3D brain MR images. We demonstrate that our approach achieved state-of-the-art registration performance while using fewer learned parameters, with good data efficiency and domain robustness.