Deep learning based geometric registration for medical images: How accurate can we get without visual features?

TL;DR

This paper explores a deep learning approach for 3D medical image registration that relies solely on geometric features, using graph convolutions and belief message passing, achieving high accuracy without visual features.

Contribution

It introduces a novel geometric feature-based registration framework combining graph convolutions and belief propagation, outperforming existing deep learning methods that depend on visual features.

Findings

Outperforms dense encoder-decoder networks in lung structure registration

Achieves high accuracy using only geometric features

Validates effectiveness on complex lung key-point graphs

Abstract

As in other areas of medical image analysis, e.g. semantic segmentation, deep learning is currently driving the development of new approaches for image registration. Multi-scale encoder-decoder network architectures achieve state-of-the-art accuracy on tasks such as intra-patient alignment of abdominal CT or brain MRI registration, especially when additional supervision, such as anatomical labels, is available. The success of these methods relies to a large extent on the outstanding ability of deep CNNs to extract descriptive visual features from the input images. In contrast to conventional methods, the explicit inclusion of geometric information plays only a minor role, if at all. In this work we take a look at an exactly opposite approach by investigating a deep learning framework for registration based solely on geometric features and optimisation. We combine graph convolutions with loopy belief message passing to enable highly accurate 3D point cloud registration. Our experimental validation is conducted on complex key-point graphs of inner lung structures, strongly outperforming dense encoder-decoder networks and other point set registration methods. Our code is publicly available at https://github.com/multimodallearning/deep-geo-reg.