Image harmonization: A review of statistical and deep learning methods for removing batch effects and evaluation metrics for effective harmonization

•Batch effects introduce significant confounding in multi-batch neuroimaging data.•Removal of batch effects is critical for reproducibility and generalizability.•We review current harmonization methods and describe common evaluation metrics.•We provide guidance to end-users on choosing an appropriat...

Full description

Saved in:
Bibliographic Details
Published in:NeuroImage (Orlando, Fla.) Vol. 274; p. 120125
Main Authors: Hu, Fengling, Chen, Andrew A., Horng, Hannah, Bashyam, Vishnu, Davatzikos, Christos, Alexander-Bloch, Aaron, Li, Mingyao, Shou, Haochang, Satterthwaite, Theodore D., Yu, Meichen, Shinohara, Russell T.
Format: Journal Article
Language:English
Published: United States Elsevier Inc 01.07.2023
Elsevier Limited
Elsevier
Subjects:
Algorithms
Benchmarking
Biomarkers
Brain - diagnostic imaging
Brain - pathology
Clinical trials
Computed tomography
Data collection
Deep Learning
Humans
Machine learning
Magnetic resonance imaging
Magnetic Resonance Imaging - methods
Medical imaging
Neuroimaging
Neurosciences
Reproducibility of Results
Scanners
Statistics
ISSN:1053-8119, 1095-9572, 1095-9572
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Batch effects introduce significant confounding in multi-batch neuroimaging data.•Removal of batch effects is critical for reproducibility and generalizability.•We review current harmonization methods and describe common evaluation metrics.•We provide guidance to end-users on choosing an appropriate harmonization method.•We provide guidance to methodologists on current limitations and future directions. Magnetic resonance imaging and computed tomography from multiple batches (e.g. sites, scanners, datasets, etc.) are increasingly used alongside complex downstream analyses to obtain new insights into the human brain. However, significant confounding due to batch-related technical variation, called batch effects, is present in this data; direct application of downstream analyses to the data may lead to biased results. Image harmonization methods seek to remove these batch effects and enable increased generalizability and reproducibility of downstream results. In this review, we describe and categorize current approaches in statistical and deep learning harmonization methods. We also describe current evaluation metrics used to assess harmonization methods and provide a standardized framework to evaluate newly-proposed methods for effective harmonization and preservation of biological information. Finally, we provide recommendations to end-users to advocate for more effective use of current methods and to methodologists to direct future efforts and accelerate development of the field.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ObjectType-Review-3
content type line 23
Asterisks indicate equal contribution to this work.
ISSN:1053-8119
1095-9572
1095-9572
DOI:10.1016/j.neuroimage.2023.120125