Cargando…

Learning the Effect of Registration Hyperparameters with HyperMorph

We introduce HyperMorph, a framework that facilitates efficient hyperparameter tuning in learning-based deformable image registration. Classical registration algorithms perform an iterative pair-wise optimization to compute a deformation field that aligns two images. Recent learning-based approaches...

Descripción completa

Detalles Bibliográficos
Autores principales: Hoopes, Andrew, Hoffmann, Malte, Greve, Douglas N., Fischl, Bruce, Guttag, John, Dalca, Adrian V.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9491317/
https://www.ncbi.nlm.nih.gov/pubmed/36147449
_version_ 1784793256726364160
author Hoopes, Andrew
Hoffmann, Malte
Greve, Douglas N.
Fischl, Bruce
Guttag, John
Dalca, Adrian V.
author_facet Hoopes, Andrew
Hoffmann, Malte
Greve, Douglas N.
Fischl, Bruce
Guttag, John
Dalca, Adrian V.
author_sort Hoopes, Andrew
collection PubMed
description We introduce HyperMorph, a framework that facilitates efficient hyperparameter tuning in learning-based deformable image registration. Classical registration algorithms perform an iterative pair-wise optimization to compute a deformation field that aligns two images. Recent learning-based approaches leverage large image datasets to learn a function that rapidly estimates a deformation for a given image pair. In both strategies, the accuracy of the resulting spatial correspondences is strongly influenced by the choice of certain hyperparameter values. However, an effective hyperparameter search consumes substantial time and human effort as it often involves training multiple models for different fixed hyperparameter values and may lead to suboptimal registration. We propose an amortized hyperparameter learning strategy to alleviate this burden by learning the impact of hyperparameters on deformation fields. We design a meta network, or hypernetwork, that predicts the parameters of a registration network for input hyperparameters, thereby comprising a single model that generates the optimal deformation field corresponding to given hyperparameter values. This strategy enables fast, high-resolution hyperparameter search at test-time, reducing the inefficiency of traditional approaches while increasing flexibility. We also demonstrate additional benefits of HyperMorph, including enhanced robustness to model initialization and the ability to rapidly identify optimal hyperparameter values specific to a dataset, image contrast, task, or even anatomical region, all without the need to retrain models. We make our code publicly available at http://hypermorph.voxelmorph.net.
format Online
Article
Text
id pubmed-9491317
institution National Center for Biotechnology Information
language English
publishDate 2022
record_format MEDLINE/PubMed
spelling pubmed-94913172022-09-21 Learning the Effect of Registration Hyperparameters with HyperMorph Hoopes, Andrew Hoffmann, Malte Greve, Douglas N. Fischl, Bruce Guttag, John Dalca, Adrian V. J Mach Learn Biomed Imaging Article We introduce HyperMorph, a framework that facilitates efficient hyperparameter tuning in learning-based deformable image registration. Classical registration algorithms perform an iterative pair-wise optimization to compute a deformation field that aligns two images. Recent learning-based approaches leverage large image datasets to learn a function that rapidly estimates a deformation for a given image pair. In both strategies, the accuracy of the resulting spatial correspondences is strongly influenced by the choice of certain hyperparameter values. However, an effective hyperparameter search consumes substantial time and human effort as it often involves training multiple models for different fixed hyperparameter values and may lead to suboptimal registration. We propose an amortized hyperparameter learning strategy to alleviate this burden by learning the impact of hyperparameters on deformation fields. We design a meta network, or hypernetwork, that predicts the parameters of a registration network for input hyperparameters, thereby comprising a single model that generates the optimal deformation field corresponding to given hyperparameter values. This strategy enables fast, high-resolution hyperparameter search at test-time, reducing the inefficiency of traditional approaches while increasing flexibility. We also demonstrate additional benefits of HyperMorph, including enhanced robustness to model initialization and the ability to rapidly identify optimal hyperparameter values specific to a dataset, image contrast, task, or even anatomical region, all without the need to retrain models. We make our code publicly available at http://hypermorph.voxelmorph.net. 2022-03 2022-04-07 /pmc/articles/PMC9491317/ /pubmed/36147449 Text en https://creativecommons.org/licenses/by/4.0/License: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/) .
spellingShingle Article
Hoopes, Andrew
Hoffmann, Malte
Greve, Douglas N.
Fischl, Bruce
Guttag, John
Dalca, Adrian V.
Learning the Effect of Registration Hyperparameters with HyperMorph
title Learning the Effect of Registration Hyperparameters with HyperMorph
title_full Learning the Effect of Registration Hyperparameters with HyperMorph
title_fullStr Learning the Effect of Registration Hyperparameters with HyperMorph
title_full_unstemmed Learning the Effect of Registration Hyperparameters with HyperMorph
title_short Learning the Effect of Registration Hyperparameters with HyperMorph
title_sort learning the effect of registration hyperparameters with hypermorph
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9491317/
https://www.ncbi.nlm.nih.gov/pubmed/36147449
work_keys_str_mv AT hoopesandrew learningtheeffectofregistrationhyperparameterswithhypermorph
AT hoffmannmalte learningtheeffectofregistrationhyperparameterswithhypermorph
AT grevedouglasn learningtheeffectofregistrationhyperparameterswithhypermorph
AT fischlbruce learningtheeffectofregistrationhyperparameterswithhypermorph
AT guttagjohn learningtheeffectofregistrationhyperparameterswithhypermorph
AT dalcaadrianv learningtheeffectofregistrationhyperparameterswithhypermorph