Group ICA for identifying biomarkers in schizophrenia: ‘Adaptive’ networks via spatially constrained ICA show more sensitivity to group differences than spatio-temporal regression

Mustafa S. Salman, Yuhui Du, Dongdong Lin, Zening Fu, Alex Fedorov, Eswar Damaraju, Jing Sui, Jiayu Chen, Andrew R. Mayer, Stefan Posse, Daniel H. Mathalon, Judith M. Ford, Theodorus Van Erp, Vince Daniel Calhoun

Research output: Contribution to journalArticle

Abstract

Brain functional networks identified from fMRI data can provide potential biomarkers for brain disorders. Group independent component analysis (GICA) is popular for extracting brain functional networks from multiple subjects. In GICA, different strategies exist for reconstructing subject-specific networks from the group-level networks. However, it is unknown whether these strategies have different sensitivities to group differences and abilities in distinguishing patients. Among GICA, spatio-temporal regression (STR) and spatially constrained ICA approaches such as group information guided ICA (GIG-ICA) can be used to propagate components (indicating networks) to a new subject that is not included in the original subjects. In this study, based on the same a priori network maps, we reconstructed subject-specific networks using these two methods separately from resting-state fMRI data of 151 schizophrenia patients (SZs) and 163 healthy controls (HCs). We investigated group differences in the estimated functional networks and the functional network connectivity (FNC) obtained by each method. The networks were also used as features in a cross-validated support vector machine (SVM) for classifying SZs and HCs. We selected features using different strategies to provide a comprehensive comparison between the two methods. GIG-ICA generally showed greater sensitivity in statistical analysis and better classification performance (accuracy 76.45 ± 8.9%, sensitivity 0.74 ± 0.11, specificity 0.79 ± 0.11) than STR (accuracy 67.45 ± 8.13%, sensitivity 0.65 ± 0.11, specificity 0.71 ± 0.11). Importantly, results were also consistent when applied to an independent dataset including 82 HCs and 82 SZs. Our work suggests that the functional networks estimated by GIG-ICA are more sensitive to group differences, and GIG-ICA is promising for identifying image-derived biomarkers of brain disease.

Original languageEnglish (US)
Article number101747
JournalNeuroImage: Clinical
Volume22
DOIs
StatePublished - Jan 1 2019
Externally publishedYes

Fingerprint

Schizophrenia
Biomarkers
Brain Diseases
Magnetic Resonance Imaging
Aptitude
Brain
Support Vector Machine
Datasets

Keywords

  • Classification
  • fMRI
  • Functional network
  • GIG-ICA
  • ICA
  • Schizophrenia
  • Spatio-temporal regression

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Neurology
  • Clinical Neurology
  • Cognitive Neuroscience

Cite this

Group ICA for identifying biomarkers in schizophrenia : ‘Adaptive’ networks via spatially constrained ICA show more sensitivity to group differences than spatio-temporal regression. / Salman, Mustafa S.; Du, Yuhui; Lin, Dongdong; Fu, Zening; Fedorov, Alex; Damaraju, Eswar; Sui, Jing; Chen, Jiayu; Mayer, Andrew R.; Posse, Stefan; Mathalon, Daniel H.; Ford, Judith M.; Van Erp, Theodorus; Calhoun, Vince Daniel.

In: NeuroImage: Clinical, Vol. 22, 101747, 01.01.2019.

Research output: Contribution to journalArticle

Salman, Mustafa S. ; Du, Yuhui ; Lin, Dongdong ; Fu, Zening ; Fedorov, Alex ; Damaraju, Eswar ; Sui, Jing ; Chen, Jiayu ; Mayer, Andrew R. ; Posse, Stefan ; Mathalon, Daniel H. ; Ford, Judith M. ; Van Erp, Theodorus ; Calhoun, Vince Daniel. / Group ICA for identifying biomarkers in schizophrenia : ‘Adaptive’ networks via spatially constrained ICA show more sensitivity to group differences than spatio-temporal regression. In: NeuroImage: Clinical. 2019 ; Vol. 22.
@article{b6cf99ec2ede4a2c8f668702c5ea7682,
title = "Group ICA for identifying biomarkers in schizophrenia: ‘Adaptive’ networks via spatially constrained ICA show more sensitivity to group differences than spatio-temporal regression",
abstract = "Brain functional networks identified from fMRI data can provide potential biomarkers for brain disorders. Group independent component analysis (GICA) is popular for extracting brain functional networks from multiple subjects. In GICA, different strategies exist for reconstructing subject-specific networks from the group-level networks. However, it is unknown whether these strategies have different sensitivities to group differences and abilities in distinguishing patients. Among GICA, spatio-temporal regression (STR) and spatially constrained ICA approaches such as group information guided ICA (GIG-ICA) can be used to propagate components (indicating networks) to a new subject that is not included in the original subjects. In this study, based on the same a priori network maps, we reconstructed subject-specific networks using these two methods separately from resting-state fMRI data of 151 schizophrenia patients (SZs) and 163 healthy controls (HCs). We investigated group differences in the estimated functional networks and the functional network connectivity (FNC) obtained by each method. The networks were also used as features in a cross-validated support vector machine (SVM) for classifying SZs and HCs. We selected features using different strategies to provide a comprehensive comparison between the two methods. GIG-ICA generally showed greater sensitivity in statistical analysis and better classification performance (accuracy 76.45 ± 8.9{\%}, sensitivity 0.74 ± 0.11, specificity 0.79 ± 0.11) than STR (accuracy 67.45 ± 8.13{\%}, sensitivity 0.65 ± 0.11, specificity 0.71 ± 0.11). Importantly, results were also consistent when applied to an independent dataset including 82 HCs and 82 SZs. Our work suggests that the functional networks estimated by GIG-ICA are more sensitive to group differences, and GIG-ICA is promising for identifying image-derived biomarkers of brain disease.",
keywords = "Classification, fMRI, Functional network, GIG-ICA, ICA, Schizophrenia, Spatio-temporal regression",
author = "Salman, {Mustafa S.} and Yuhui Du and Dongdong Lin and Zening Fu and Alex Fedorov and Eswar Damaraju and Jing Sui and Jiayu Chen and Mayer, {Andrew R.} and Stefan Posse and Mathalon, {Daniel H.} and Ford, {Judith M.} and {Van Erp}, Theodorus and Calhoun, {Vince Daniel}",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.nicl.2019.101747",
language = "English (US)",
volume = "22",
journal = "NeuroImage: Clinical",
issn = "2213-1582",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Group ICA for identifying biomarkers in schizophrenia

T2 - ‘Adaptive’ networks via spatially constrained ICA show more sensitivity to group differences than spatio-temporal regression

AU - Salman, Mustafa S.

AU - Du, Yuhui

AU - Lin, Dongdong

AU - Fu, Zening

AU - Fedorov, Alex

AU - Damaraju, Eswar

AU - Sui, Jing

AU - Chen, Jiayu

AU - Mayer, Andrew R.

AU - Posse, Stefan

AU - Mathalon, Daniel H.

AU - Ford, Judith M.

AU - Van Erp, Theodorus

AU - Calhoun, Vince Daniel

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Brain functional networks identified from fMRI data can provide potential biomarkers for brain disorders. Group independent component analysis (GICA) is popular for extracting brain functional networks from multiple subjects. In GICA, different strategies exist for reconstructing subject-specific networks from the group-level networks. However, it is unknown whether these strategies have different sensitivities to group differences and abilities in distinguishing patients. Among GICA, spatio-temporal regression (STR) and spatially constrained ICA approaches such as group information guided ICA (GIG-ICA) can be used to propagate components (indicating networks) to a new subject that is not included in the original subjects. In this study, based on the same a priori network maps, we reconstructed subject-specific networks using these two methods separately from resting-state fMRI data of 151 schizophrenia patients (SZs) and 163 healthy controls (HCs). We investigated group differences in the estimated functional networks and the functional network connectivity (FNC) obtained by each method. The networks were also used as features in a cross-validated support vector machine (SVM) for classifying SZs and HCs. We selected features using different strategies to provide a comprehensive comparison between the two methods. GIG-ICA generally showed greater sensitivity in statistical analysis and better classification performance (accuracy 76.45 ± 8.9%, sensitivity 0.74 ± 0.11, specificity 0.79 ± 0.11) than STR (accuracy 67.45 ± 8.13%, sensitivity 0.65 ± 0.11, specificity 0.71 ± 0.11). Importantly, results were also consistent when applied to an independent dataset including 82 HCs and 82 SZs. Our work suggests that the functional networks estimated by GIG-ICA are more sensitive to group differences, and GIG-ICA is promising for identifying image-derived biomarkers of brain disease.

AB - Brain functional networks identified from fMRI data can provide potential biomarkers for brain disorders. Group independent component analysis (GICA) is popular for extracting brain functional networks from multiple subjects. In GICA, different strategies exist for reconstructing subject-specific networks from the group-level networks. However, it is unknown whether these strategies have different sensitivities to group differences and abilities in distinguishing patients. Among GICA, spatio-temporal regression (STR) and spatially constrained ICA approaches such as group information guided ICA (GIG-ICA) can be used to propagate components (indicating networks) to a new subject that is not included in the original subjects. In this study, based on the same a priori network maps, we reconstructed subject-specific networks using these two methods separately from resting-state fMRI data of 151 schizophrenia patients (SZs) and 163 healthy controls (HCs). We investigated group differences in the estimated functional networks and the functional network connectivity (FNC) obtained by each method. The networks were also used as features in a cross-validated support vector machine (SVM) for classifying SZs and HCs. We selected features using different strategies to provide a comprehensive comparison between the two methods. GIG-ICA generally showed greater sensitivity in statistical analysis and better classification performance (accuracy 76.45 ± 8.9%, sensitivity 0.74 ± 0.11, specificity 0.79 ± 0.11) than STR (accuracy 67.45 ± 8.13%, sensitivity 0.65 ± 0.11, specificity 0.71 ± 0.11). Importantly, results were also consistent when applied to an independent dataset including 82 HCs and 82 SZs. Our work suggests that the functional networks estimated by GIG-ICA are more sensitive to group differences, and GIG-ICA is promising for identifying image-derived biomarkers of brain disease.

KW - Classification

KW - fMRI

KW - Functional network

KW - GIG-ICA

KW - ICA

KW - Schizophrenia

KW - Spatio-temporal regression

UR - http://www.scopus.com/inward/record.url?scp=85063318600&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063318600&partnerID=8YFLogxK

U2 - 10.1016/j.nicl.2019.101747

DO - 10.1016/j.nicl.2019.101747

M3 - Article

C2 - 30921608

AN - SCOPUS:85063318600

VL - 22

JO - NeuroImage: Clinical

JF - NeuroImage: Clinical

SN - 2213-1582

M1 - 101747

ER -