Deep learning in medical image registration: introduction and survey

TL;DR

This survey reviews deep learning techniques for medical image registration, covering algorithms, transformations, datasets, evaluation metrics, and applications, while discussing future directions like transformer-based models.

Contribution

It provides a comprehensive overview of deep learning methods in medical image registration, highlighting recent advances, challenges, and future research directions.

Findings

Deep learning methods improve registration accuracy and efficiency.

Transformers are emerging as promising tools for future research.

Various datasets and metrics are used for evaluating registration performance.

Abstract

Image registration (IR) is a process that deforms images to align them with respect to a reference space, making it easier for medical practitioners to examine various medical images in a standardized reference frame, such as having the same rotation and scale. This document introduces image registration using a simple numeric example. It provides a definition of image registration along with a space-oriented symbolic representation. This review covers various aspects of image transformations, including affine, deformable, invertible, and bidirectional transformations, as well as medical image registration algorithms such as Voxelmorph, Demons, SyN, Iterative Closest Point, and SynthMorph. It also explores atlas-based registration and multistage image registration techniques, including coarse-fine and pyramid approaches. Furthermore, this survey paper discusses medical image registration taxonomies, datasets, evaluation measures, such as correlation-based metrics, segmentation-based metrics, processing time, and model size. It also explores applications in image-guided surgery, motion tracking, and tumor diagnosis. Finally, the document addresses future research directions, including the further development of transformers.