TY - JOUR
T1 - Multisite generalizability of schizophrenia diagnosis classification based on functional brain connectivity
AU - Orban, Pierre
AU - Dansereau, Christian
AU - Desbois, Laurence
AU - Mongeau-Pérusse, Violaine
AU - Giguère, Charles Édouard
AU - Nguyen, Hien
AU - Mendrek, Adrianna
AU - Stip, Emmanuel
AU - Bellec, Pierre
N1 - Funding Information:
Data from one study (emotional memory task) were collected thanks to grants from the Canadian Institutes of Health Research, Gender and Health Institute to Dr. Mendrek (CIHR grant number 200603MOP-158161-GSH-CFCA-130656 ). Data from 4 other studies were accessed through the SchizConnect platform ( http://schizconnect.org ). As such, the investigators within SchizConnect contributed to the design and implementation of SchizConnect and/or provided data but did not participate in analysis or writing of this report. Funding of the SchizConnect project was provided by NIMH cooperative agreement 1U01 MH097435 . SchizConnect enabled access to the following data repository: the COllaborative Informatics and Neuroimaging Suite Data Exchange tool (COINS; http://coins.mrn.org/dx ). Data from one study (resting-state) was collected at the Mind Research Network and funded by a Center of Biomedical Research Excellence (COBRE) grant 5P20RR021938/P20GM103472 from the NIH to Dr. Vince Calhoun. Data from two studies (oddball and Sternberg item recognition paradigm tasks) were obtained from the Mind Clinical Imaging Consortium database. The MCIC project was supported by the Department of Energy under award number DE-FG02-08ER6458 . MCIC is the result of efforts of co-investigators from University of Iowa, University of Minnesota, University of New Mexico and Massachusetts General Hospital. Data from a fourth study (N-back task) were obtained from the Neuromorphometry by Computer Algorithm Chicago (NMorphCH) dataset ( http://nunda.northwestern.edu/nunda/data/projects/NMorphCH ) As such, the investigators within NMorphCH contributed to the design and implementation of NMorphCH and/or provided data but did not participate in analysis or writing of this report. The NMorphCH project was funded by NIMH grant RO1 MH056584 . The last study (task-switching) was obtained through the OpenFMRI project ( http://openfmri.org ) from the Consortium for Neuropsychiatric Phenomics (CNP), which was supported by NIH Roadmap for Medical Research grants UL1-DE019580 , RL1MH083268 , RL1MH083269 , RL1DA024853 , RL1MH083270 , RL1LM009833 , PL1MH083271 , and PL1NS062410 . Data analysis was supported by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC grant number # 436141 ) to PB. CD is supported by a bursary from the Lemaire foundation .
Funding Information:
Data from one study (emotional memory task) were collected thanks to grants from the Canadian Institutes of Health Research, Gender and Health Institute to Dr. Mendrek (CIHR grant number 200603MOP-158161-GSH-CFCA-130656). Data from 4 other studies were accessed through the SchizConnect platform (http://schizconnect.org). As such, the investigators within SchizConnect contributed to the design and implementation of SchizConnect and/or provided data but did not participate in analysis or writing of this report. Funding of the SchizConnect project was provided by NIMH cooperative agreement 1U01 MH097435. SchizConnect enabled access to the following data repository: the COllaborative Informatics and Neuroimaging Suite Data Exchange tool (COINS; http://coins.mrn.org/dx). Data from one study (resting-state) was collected at the Mind Research Network and funded by a Center of Biomedical Research Excellence (COBRE) grant 5P20RR021938/P20GM103472 from the NIH to Dr. Vince Calhoun. Data from two studies (oddball and Sternberg item recognition paradigm tasks) were obtained from the Mind Clinical Imaging Consortium database. The MCIC project was supported by the Department of Energy under award number DE-FG02-08ER6458. MCIC is the result of efforts of co-investigators from University of Iowa, University of Minnesota, University of New Mexico and Massachusetts General Hospital. Data from a fourth study (N-back task) were obtained from the Neuromorphometry by Computer Algorithm Chicago (NMorphCH) dataset (http://nunda.northwestern.edu/nunda/data/projects/NMorphCH) As such, the investigators within NMorphCH contributed to the design and implementation of NMorphCH and/or provided data but did not participate in analysis or writing of this report. The NMorphCH project was funded by NIMH grant RO1 MH056584. The last study (task-switching) was obtained through the OpenFMRI project (http://openfmri.org) from the Consortium for Neuropsychiatric Phenomics (CNP), which was supported by NIH Roadmap for Medical Research grants UL1-DE019580, RL1MH083268, RL1MH083269, RL1DA024853, RL1MH083270, RL1LM009833, PL1MH083271, and PL1NS062410. Data analysis was supported by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC grant number #436141) to PB. CD is supported by a bursary from the Lemaire foundation.
Publisher Copyright:
© 2017
PY - 2018/2
Y1 - 2018/2
N2 - Our objective was to assess the generalizability, across sites and cognitive contexts, of schizophrenia classification based on functional brain connectivity. We tested different training-test scenarios combining fMRI data from 191 schizophrenia patients and 191 matched healthy controls obtained at 6 scanning sites and under different task conditions. Diagnosis classification accuracy generalized well to a novel site and cognitive context provided data from multiple sites were used for classifier training. By contrast, lower classification accuracy was achieved when data from a single distinct site was used for training. These findings indicate that it is beneficial to use multisite data to train fMRI-based classifiers intended for large-scale use in the clinical realm.
AB - Our objective was to assess the generalizability, across sites and cognitive contexts, of schizophrenia classification based on functional brain connectivity. We tested different training-test scenarios combining fMRI data from 191 schizophrenia patients and 191 matched healthy controls obtained at 6 scanning sites and under different task conditions. Diagnosis classification accuracy generalized well to a novel site and cognitive context provided data from multiple sites were used for classifier training. By contrast, lower classification accuracy was achieved when data from a single distinct site was used for training. These findings indicate that it is beneficial to use multisite data to train fMRI-based classifiers intended for large-scale use in the clinical realm.
KW - Classification
KW - Cognition
KW - Machine learning
KW - Multisite
KW - Schizophrenia
KW - fMRI
UR - http://www.scopus.com/inward/record.url?scp=85020280347&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85020280347&partnerID=8YFLogxK
U2 - 10.1016/j.schres.2017.05.027
DO - 10.1016/j.schres.2017.05.027
M3 - Article
C2 - 28601499
AN - SCOPUS:85020280347
SN - 0920-9964
VL - 192
SP - 167
EP - 171
JO - Schizophrenia Research
JF - Schizophrenia Research
ER -