Dense Deformation Network for High Resolution Tissue Cleared Image Registration

TL;DR

This paper introduces a densely connected convolutional network for high-resolution tissue image registration that outperforms existing deep learning methods in accuracy and speed, especially at higher resolutions.

Contribution

The paper proposes a novel dense network architecture with single-stage downsampling and dense connections, improving registration accuracy and resolution handling without ground-truth labels.

Findings

Achieves comparable performance to state-of-the-art methods.

Outperforms VoxelMorph in accuracy and resolution handling.

Registers brain images in one minute, much faster than traditional methods.

Abstract

The recent application of deep learning in various areas of medical image analysis has brought excellent performance gains. In particular, technologies based on deep learning in medical image registration can outperform traditional optimisation-based registration algorithms both in registration time and accuracy. However, the U-net based architectures used in most of the image registration frameworks downscale the data, which removes global information and affects the deformation. In this paper, we present a densely connected convolutional architecture for deformable image registration. Our proposed dense network downsizes data only in one stage and have dense connections instead of the skip connections in U-net architecture. The training of the network is unsupervised and does not require ground-truth deformation or any synthetic deformation as a label. The proposed architecture is trained and tested on two different versions of tissue-cleared data, at 10\% and 25\% resolution of the original single-cell-resolution dataset. We demonstrate comparable registration performance to state-of-the-art registration methods and superior performance to the deep-learning based VoxelMorph method in terms of accuracy and increased resolution handling ability. In both resolutions, the proposed DenseDeformation network outperforms VoxelMorph in registration accuracy. Importantly, it can register brains in one minute where conventional methods can take hours at 25\% resolution.