Detecting schizophrenia at the level of the individual: relative diagnostic value of whole-brain images, connectome-wide functional connectivity and graph-based metrics.
functional connectivity
graph theoretical analysis
machine learning
neuroimaging
schizophrenia.
Journal
Psychological medicine
ISSN: 1469-8978
Titre abrégé: Psychol Med
Pays: England
ID NLM: 1254142
Informations de publication
Date de publication:
08 2020
08 2020
Historique:
pubmed:
9
8
2019
medline:
16
6
2021
entrez:
9
8
2019
Statut:
ppublish
Résumé
Previous studies using resting-state functional neuroimaging have revealed alterations in whole-brain images, connectome-wide functional connectivity and graph-based metrics in groups of patients with schizophrenia relative to groups of healthy controls. However, it is unclear which of these measures best captures the neural correlates of this disorder at the level of the individual patient. Here we investigated the relative diagnostic value of these measures. A total of 295 patients with schizophrenia and 452 healthy controls were investigated using resting-state functional Magnetic Resonance Imaging at five research centres. Connectome-wide functional networks were constructed by thresholding correlation matrices of 90 brain regions, and their topological properties were analyzed using graph theory-based methods. Single-subject classification was performed using three machine learning (ML) approaches associated with varying degrees of complexity and abstraction, namely logistic regression, support vector machine and deep learning technology. Connectome-wide functional connectivity allowed single-subject classification of patients and controls with higher accuracy (average: 81%) than both whole-brain images (average: 53%) and graph-based metrics (average: 69%). Classification based on connectome-wide functional connectivity was driven by a distributed bilateral network including the thalamus and temporal regions. These results were replicated across the three employed ML approaches. Connectome-wide functional connectivity permits differentiation of patients with schizophrenia from healthy controls at single-subject level with greater accuracy; this pattern of results is consistent with the 'dysconnectivity hypothesis' of schizophrenia, which states that the neural basis of the disorder is best understood in terms of system-level functional connectivity alterations.
Sections du résumé
BACKGROUND
Previous studies using resting-state functional neuroimaging have revealed alterations in whole-brain images, connectome-wide functional connectivity and graph-based metrics in groups of patients with schizophrenia relative to groups of healthy controls. However, it is unclear which of these measures best captures the neural correlates of this disorder at the level of the individual patient.
METHODS
Here we investigated the relative diagnostic value of these measures. A total of 295 patients with schizophrenia and 452 healthy controls were investigated using resting-state functional Magnetic Resonance Imaging at five research centres. Connectome-wide functional networks were constructed by thresholding correlation matrices of 90 brain regions, and their topological properties were analyzed using graph theory-based methods. Single-subject classification was performed using three machine learning (ML) approaches associated with varying degrees of complexity and abstraction, namely logistic regression, support vector machine and deep learning technology.
RESULTS
Connectome-wide functional connectivity allowed single-subject classification of patients and controls with higher accuracy (average: 81%) than both whole-brain images (average: 53%) and graph-based metrics (average: 69%). Classification based on connectome-wide functional connectivity was driven by a distributed bilateral network including the thalamus and temporal regions.
CONCLUSION
These results were replicated across the three employed ML approaches. Connectome-wide functional connectivity permits differentiation of patients with schizophrenia from healthy controls at single-subject level with greater accuracy; this pattern of results is consistent with the 'dysconnectivity hypothesis' of schizophrenia, which states that the neural basis of the disorder is best understood in terms of system-level functional connectivity alterations.
Identifiants
pubmed: 31391132
pii: S0033291719001934
doi: 10.1017/S0033291719001934
pmc: PMC7477363
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
1852-1861Subventions
Organisme : Wellcome Trust
Pays : United Kingdom
Organisme : Wellcome Trust
ID : 208519/Z/17/Z
Pays : United Kingdom
Références
Nature. 1998 Jun 4;393(6684):440-2
pubmed: 9623998
Neuropsychiatr Dis Treat. 2017 Feb 14;13:467-475
pubmed: 28243099
J Neural Transm (Vienna). 2015 Jan;122(1):123-34
pubmed: 25037587
Biol Psychiatry. 2016 Jun 15;79(12):1016-25
pubmed: 26248537
Neuroimage. 2016 Jan 1;124(Pt A):127-146
pubmed: 25987366
Front Hum Neurosci. 2015 Jun 30;9:386
pubmed: 26175682
Schizophr Res. 2016 Oct;176(2-3):83-94
pubmed: 27450778
BMC Psychiatry. 2016 Jul 11;16:228
pubmed: 27400680
Neuroimage. 2014 Aug 15;97:117-26
pubmed: 24736181
Med Image Anal. 2017 Jan;35:18-31
pubmed: 27310171
Hum Brain Mapp. 2015 Jul;36(7):2629-43
pubmed: 25832856
Neuroinformatics. 2015 Jan;13(1):31-46
pubmed: 25048627
Neurosci Biobehav Rev. 2012 Apr;36(4):1140-52
pubmed: 22305994
Cereb Cortex. 2007 Oct;17(10):2407-19
pubmed: 17204824
J Neurosci. 2010 Jul 14;30(28):9477-87
pubmed: 20631176
Brain Imaging Behav. 2016 Sep;10(3):799-817
pubmed: 26363784
Annu Rev Clin Psychol. 2011;7:113-40
pubmed: 21128784
Schizophr Res. 2015 Oct;168(1-2):345-52
pubmed: 26299706
Neurosci Bull. 2017 Feb;33(1):73-84
pubmed: 27995564
Psychol Med. 2013 Dec;43(12):2547-62
pubmed: 23507081
Nature. 2015 May 28;521(7553):436-44
pubmed: 26017442
Neuroinformatics. 2014 Apr;12(2):229-44
pubmed: 24013948
Proc Natl Acad Sci U S A. 2010 Mar 9;107(10):4734-9
pubmed: 20176931
J Psychiatry Neurosci. 2018 Nov 01;43(6):427
pubmed: 30375837
J Neurosci. 2011 Jan 26;31(4):1204-12
pubmed: 21273405
Phys Rev E Stat Nonlin Soft Matter Phys. 2003 Feb;67(2 Pt 2):026126
pubmed: 12636767
Neuroimage. 2014 Sep;98:405-15
pubmed: 24791746
Cereb Cortex. 2005 Sep;15(9):1332-42
pubmed: 15635061
Schizophr Bull. 2017 Jul 1;43(4):914-924
pubmed: 27872268
Nat Rev Neurosci. 2009 Mar;10(3):186-98
pubmed: 19190637
Am J Psychiatry. 2012 Oct;169(10):1092-9
pubmed: 23032387
Neuroimage. 2012 Oct 1;62(4):2296-314
pubmed: 22387165
Sci Rep. 2016 Dec 12;6:38897
pubmed: 27941946
Inf Process Med Imaging. 2003 Jul;18:330-41
pubmed: 15344469
Neuroimage. 2010 Feb 15;49(4):3110-21
pubmed: 19931396
Cortex. 2015 Feb;63:55-67
pubmed: 25243989
Schizophr Bull. 2011 May;37(3):640-50
pubmed: 19920062
Can J Psychiatry. 2014 Jan;59(1):13-7
pubmed: 24444319
Clin Neurophysiol. 2015 Nov;126(11):2132-41
pubmed: 25907414
Biol Psychiatry. 2011 Aug 15;70(4):334-42
pubmed: 21791259
J Neurophysiol. 2009 Jun;101(6):3270-83
pubmed: 19339462
Hum Brain Mapp. 2015 Sep;36(9):3677-86
pubmed: 26096541
Trends Cogn Sci. 2013 Dec;17(12):641-7
pubmed: 24238779
Radiology. 2015 Sep;276(3):818-27
pubmed: 25848901
Cereb Cortex. 2015 Sep;25(9):3046-56
pubmed: 24836690
Neurosci Biobehav Rev. 2017 Mar;74(Pt A):58-75
pubmed: 28087243
Phys Rev Lett. 2001 Nov 5;87(19):198701
pubmed: 11690461
Neurosci Biobehav Rev. 2011 Apr;35(5):1110-24
pubmed: 21115039
JAMA Psychiatry. 2015 Sep;72(9):882-91
pubmed: 26267151
PLoS One. 2012;7(5):e37828
pubmed: 22666397