A guided multiverse study of neuroimaging analyses

Jessica Dafflon; Pedro F Da Costa; František Váša; Ricardo Pio Monti; Danilo Bzdok; Peter J Hellyer; Federico Turkheimer; Jonathan Smallwood; Emily Jones; Robert Leech

doi:10.1038/s41467-022-31347-8

A guided multiverse study of neuroimaging analyses

Jessica Dafflon, Pedro F Da Costa, František Váša, Ricardo Pio Monti, Danilo Bzdok, Peter J Hellyer, Federico Turkheimer, Jonathan Smallwood, Emily Jones, Robert Leech

King's College London
Centre for Neuroimaging Sciences
Univ London, Birkbeck University London, University College London, University of London Royal Veterinary College, University of London, St Georges University London, Imperial College London, Kings College London, Royal Holloway University London, Queen Mary University London, Div Populat Hlth Sci & Educ
McGill Univ, McGill University, Royal Victoria Hospital, Royal Victoria Hosp, Ctr Hlth, Dept Clin Epidemiol
Queen's University Cancer Research Institute
Department of Computer Science and Information Systems, Birkbeck College

Research output: Contribution to journal › Article › peer-review

5 Citations (Scopus)

Abstract

For most neuroimaging questions the range of possible analytic choices makes it unclear how to evaluate conclusions from any single analytic method. One possible way to address this issue is to evaluate all possible analyses using a multiverse approach, however, this can be computationally challenging and sequential analyses on the same data can compromise predictive power. Here, we establish how active learning on a low-dimensional space capturing the inter-relationships between pipelines can efficiently approximate the full spectrum of analyses. This approach balances the benefits of a multiverse analysis without incurring the cost on computational and predictive power. We illustrate this approach with two functional MRI datasets (predicting brain age and autism diagnosis) demonstrating how a multiverse of analyses can be efficiently navigated and mapped out using active learning. Furthermore, our presented approach not only identifies the subset of analysis techniques that are best able to predict age or classify individuals with autism spectrum disorder and healthy controls, but it also allows the relationships between analyses to be quantified.

Original language	English
Article number	3758
Pages (from-to)	3758
Journal	Nature Communications
Volume	13
Issue number	1
DOIs	https://doi.org/10.1038/s41467-022-31347-8
Publication status	Published - 29 Jun 2022

Keywords

Autism Spectrum Disorder/diagnostic imaging
Brain/diagnostic imaging
Humans
Magnetic Resonance Imaging/methods
Neuroimaging/methods

Access to Document

10.1038/s41467-022-31347-8

Cite this

@article{ca54946f80d94abba2a552b4bf47b1c2,

title = "A guided multiverse study of neuroimaging analyses",

abstract = "For most neuroimaging questions the range of possible analytic choices makes it unclear how to evaluate conclusions from any single analytic method. One possible way to address this issue is to evaluate all possible analyses using a multiverse approach, however, this can be computationally challenging and sequential analyses on the same data can compromise predictive power. Here, we establish how active learning on a low-dimensional space capturing the inter-relationships between pipelines can efficiently approximate the full spectrum of analyses. This approach balances the benefits of a multiverse analysis without incurring the cost on computational and predictive power. We illustrate this approach with two functional MRI datasets (predicting brain age and autism diagnosis) demonstrating how a multiverse of analyses can be efficiently navigated and mapped out using active learning. Furthermore, our presented approach not only identifies the subset of analysis techniques that are best able to predict age or classify individuals with autism spectrum disorder and healthy controls, but it also allows the relationships between analyses to be quantified.",

keywords = "Autism Spectrum Disorder/diagnostic imaging, Brain/diagnostic imaging, Humans, Magnetic Resonance Imaging/methods, Neuroimaging/methods",

author = "Jessica Dafflon and {F Da Costa}, Pedro and Franti{\v s}ek V{\'a}{\v s}a and Monti, {Ricardo Pio} and Danilo Bzdok and Hellyer, {Peter J} and Federico Turkheimer and Jonathan Smallwood and Emily Jones and Robert Leech",

note = "Funding Information: We would like to thank the ABIDE initiative for sharing the dataset (http://fcon_1000.projects.nitrc.org/indi/abide/) and the Neuro Bureau Preprocessing Initiative for proving the pre-processed ABIDE dataset ( https://preprocessed-connectomes-project.org/abide/ ). R.L. was funded by the Medical Research Council (Ref: MR/R005370/1), Wellcome/EPSRC Centre for Medical Engineering (Ref: WT 203148/Z/16/Z), Simons Foundation (SFG640710) and support from the Data to Early Diagnosis and Precision Medicine Industrial Strategy Challenge Fund, UK Research and Innovation (UKRI); J.D. was funded by the King{\textquoteright}s College London & Imperial College London EPSRC Centre for Doctoral Training in Medical Imaging (EP/L015226/1). The authors would also like to acknowledge support from the Data to Early Diagnosis and Precision Medicine Industrial Strategy Challenge Fund, UK Research and Innovation (UKRI). Publisher Copyright: {\textcopyright} 2022, The Author(s).",

year = "2022",

month = jun,

day = "29",

doi = "10.1038/s41467-022-31347-8",

language = "English",

volume = "13",

pages = "3758",

journal = "Nature Communications",

issn = "2041-1723",

publisher = "Nature Publishing Group",

number = "1",

}

TY - JOUR

T1 - A guided multiverse study of neuroimaging analyses

AU - Dafflon, Jessica

AU - F Da Costa, Pedro

AU - Váša, František

AU - Monti, Ricardo Pio

AU - Bzdok, Danilo

AU - Hellyer, Peter J

AU - Turkheimer, Federico

AU - Smallwood, Jonathan

AU - Jones, Emily

AU - Leech, Robert

N1 - Funding Information: We would like to thank the ABIDE initiative for sharing the dataset (http://fcon_1000.projects.nitrc.org/indi/abide/) and the Neuro Bureau Preprocessing Initiative for proving the pre-processed ABIDE dataset ( https://preprocessed-connectomes-project.org/abide/ ). R.L. was funded by the Medical Research Council (Ref: MR/R005370/1), Wellcome/EPSRC Centre for Medical Engineering (Ref: WT 203148/Z/16/Z), Simons Foundation (SFG640710) and support from the Data to Early Diagnosis and Precision Medicine Industrial Strategy Challenge Fund, UK Research and Innovation (UKRI); J.D. was funded by the King’s College London & Imperial College London EPSRC Centre for Doctoral Training in Medical Imaging (EP/L015226/1). The authors would also like to acknowledge support from the Data to Early Diagnosis and Precision Medicine Industrial Strategy Challenge Fund, UK Research and Innovation (UKRI). Publisher Copyright: © 2022, The Author(s).

PY - 2022/6/29

Y1 - 2022/6/29

N2 - For most neuroimaging questions the range of possible analytic choices makes it unclear how to evaluate conclusions from any single analytic method. One possible way to address this issue is to evaluate all possible analyses using a multiverse approach, however, this can be computationally challenging and sequential analyses on the same data can compromise predictive power. Here, we establish how active learning on a low-dimensional space capturing the inter-relationships between pipelines can efficiently approximate the full spectrum of analyses. This approach balances the benefits of a multiverse analysis without incurring the cost on computational and predictive power. We illustrate this approach with two functional MRI datasets (predicting brain age and autism diagnosis) demonstrating how a multiverse of analyses can be efficiently navigated and mapped out using active learning. Furthermore, our presented approach not only identifies the subset of analysis techniques that are best able to predict age or classify individuals with autism spectrum disorder and healthy controls, but it also allows the relationships between analyses to be quantified.

AB - For most neuroimaging questions the range of possible analytic choices makes it unclear how to evaluate conclusions from any single analytic method. One possible way to address this issue is to evaluate all possible analyses using a multiverse approach, however, this can be computationally challenging and sequential analyses on the same data can compromise predictive power. Here, we establish how active learning on a low-dimensional space capturing the inter-relationships between pipelines can efficiently approximate the full spectrum of analyses. This approach balances the benefits of a multiverse analysis without incurring the cost on computational and predictive power. We illustrate this approach with two functional MRI datasets (predicting brain age and autism diagnosis) demonstrating how a multiverse of analyses can be efficiently navigated and mapped out using active learning. Furthermore, our presented approach not only identifies the subset of analysis techniques that are best able to predict age or classify individuals with autism spectrum disorder and healthy controls, but it also allows the relationships between analyses to be quantified.

KW - Autism Spectrum Disorder/diagnostic imaging

KW - Brain/diagnostic imaging

KW - Humans

KW - Magnetic Resonance Imaging/methods

KW - Neuroimaging/methods

UR - http://www.scopus.com/inward/record.url?scp=85133140416&partnerID=8YFLogxK

U2 - 10.1038/s41467-022-31347-8

DO - 10.1038/s41467-022-31347-8

M3 - Article

C2 - 35768409

SN - 2041-1723

VL - 13

SP - 3758

JO - Nature Communications

JF - Nature Communications

IS - 1

M1 - 3758

ER -

A guided multiverse study of neuroimaging analyses

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this