To Learn or Not to Learn Features for Deformable Registration?

TL;DR

This paper investigates whether features learned through deep learning frameworks improve deformable registration performance and finds that deep learning features perform comparably to traditional SSC features across datasets.

Contribution

It provides a comparative analysis of learned deep learning features versus traditional features in deformable registration, highlighting the conditions under which DL features are beneficial.

Findings

DL features perform comparably to SSC features in registration tasks

Performance stability of DL features across datasets

Low-level features are less effective than learned DL features

Abstract

Feature-based registration has been popular with a variety of features ranging from voxel intensity to Self-Similarity Context (SSC). In this paper, we examine the question on how features learnt using various Deep Learning (DL) frameworks can be used for deformable registration and whether this feature learning is necessary or not. We investigate the use of features learned by different DL methods in the current state-of-the-art discrete registration framework and analyze its performance on 2 publicly available datasets. We draw insights into the type of DL framework useful for feature learning and the impact, if any, of the complexity of different DL models and brain parcellation methods on the performance of discrete registration. Our results indicate that the registration performance with DL features and SSC are comparable and stable across datasets whereas this does not hold for low level features.