3D Convolutional Neural Networks Image Registration Based on Efficient Supervised Learning from Artificial Deformations

TL;DR

This paper introduces RegNet, a supervised 3D CNN-based method for nonrigid chest CT image registration trained on artificially generated displacement fields, achieving accurate results efficiently.

Contribution

It presents a novel supervised learning approach using artificial DVFs for 3D image registration, with multiple architectures and a multi-stage scheme for improved accuracy.

Findings

Achieved target registration errors of 2.32 mm and 1.86 mm on two datasets.

Demonstrated inference time of approximately 2.2 seconds.

Validated effectiveness across multiple chest CT databases.

Abstract

We propose a supervised nonrigid image registration method, trained using artificial displacement vector fields (DVF), for which we propose and compare three network architectures. The artificial DVFs allow training in a fully supervised and voxel-wise dense manner, but without the cost usually associated with the creation of densely labeled data. We propose a scheme to artificially generate DVFs, and for chest CT registration augment these with simulated respiratory motion. The proposed architectures are embedded in a multi-stage approach, to increase the capture range of the proposed networks in order to more accurately predict larger displacements. The proposed method, RegNet, is evaluated on multiple databases of chest CT scans and achieved a target registration error of 2.32 $\pm$ 5.33 mm and 1.86 $\pm$ 2.12 mm on SPREAD and DIR-Lab-4DCT studies, respectively. The average inference time of RegNet with two stages is about 2.2 s.