Home > Research > Publications & Outputs > Can we accurately classify schizophrenia patien...

Electronic data

  • SCHRES-D-17-00102R1%281%29

    Rights statement: This is the author’s version of a work that was accepted for publication in Schizophrenia Research. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Schizophrenia Research 214, 2018 DOI: 10.1016/j.schres.2017.11.038

    Accepted author manuscript, 1.11 MB, PDF document

    Available under license: CC BY-NC-ND: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

Links

Text available via DOI:

View graph of relations

Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning?: A multi-method and multi-dataset study

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Published

Standard

Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning? A multi-method and multi-dataset study. / Winterburn, Julie L; Voineskos, Aristotle N; Devenyi, Gabriel A et al.
In: Schizophrenia Research, Vol. 214, 01.12.2019, p. 3-10.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

Winterburn, JL, Voineskos, AN, Devenyi, GA, Plitman, E, de la Fuente-Sandoval, C, Bhagwat, N, Graff-Guerrero, A, Knight, J & Chakravarty, MM 2019, 'Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning? A multi-method and multi-dataset study', Schizophrenia Research, vol. 214, pp. 3-10. https://doi.org/10.1016/j.schres.2017.11.038

APA

Winterburn, J. L., Voineskos, A. N., Devenyi, G. A., Plitman, E., de la Fuente-Sandoval, C., Bhagwat, N., Graff-Guerrero, A., Knight, J., & Chakravarty, M. M. (2019). Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning? A multi-method and multi-dataset study. Schizophrenia Research, 214, 3-10. https://doi.org/10.1016/j.schres.2017.11.038

Vancouver

Winterburn JL, Voineskos AN, Devenyi GA, Plitman E, de la Fuente-Sandoval C, Bhagwat N et al. Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning? A multi-method and multi-dataset study. Schizophrenia Research. 2019 Dec 1;214:3-10. Epub 2017 Dec 21. doi: 10.1016/j.schres.2017.11.038

Author

Winterburn, Julie L ; Voineskos, Aristotle N ; Devenyi, Gabriel A et al. / Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning? A multi-method and multi-dataset study. In: Schizophrenia Research. 2019 ; Vol. 214. pp. 3-10.

Bibtex

@article{74bd9a2b5ff54ffba248ed18d4e96825,
title = "Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning?: A multi-method and multi-dataset study",
abstract = "Machine learning is a powerful tool that has previously been used to classify schizophrenia (SZ) patients from healthy controls (HC) using magnetic resonance images. Each study, however, uses different datasets, classification algorithms, and validation techniques. Here, we perform a critical appraisal of the accuracy of machine learning methodologies used in SZ/HC classifications studies by comparing three machine learning algorithms (logistic regression [LR], support vector machines [SVMs], and linear discriminant analysis [LDA]) on three independent datasets (435 subjects total) using two tissue density estimates and cortical thickness (CT). Performance is assessed using 10-fold cross-validation, as well as a held-out validation set. Classification using CT outperformed tissue densities, but there was no clear effect of dataset. LR, SVMs, and LDA each yielded the highest accuracies for a different feature set and validation paradigm, but most accuracies were between 55 and 70%, well below previously reported values. The highest accuracy achieved was 73.5% using CT data and an SVM. Taken together, these results illustrate some of the obstacles to constructing effective disease classifiers, and suggest that tissue densities and CT may not be sufficiently sensitive for SZ/HC classification given current available methodologies and sample sizes.",
keywords = "Structural magnetic resonance imaging, Machine learning, Classification, Schizophrenia, Voxel-based morphometry, Cortical thickness",
author = "Winterburn, {Julie L} and Voineskos, {Aristotle N} and Devenyi, {Gabriel A} and Eric Plitman and {de la Fuente-Sandoval}, Camilo and Nikhil Bhagwat and Ariel Graff-Guerrero and Jo Knight and Chakravarty, {M Mallar}",
note = "This is the author{\textquoteright}s version of a work that was accepted for publication in Schizophrenia Research. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Schizophrenia Research 214, 2018 DOI: 10.1016/j.schres.2017.11.038",
year = "2019",
month = dec,
day = "1",
doi = "10.1016/j.schres.2017.11.038",
language = "English",
volume = "214",
pages = "3--10",
journal = "Schizophrenia Research",
issn = "0920-9964",
publisher = "ELSEVIER SCIENCE BV",

}

RIS

TY - JOUR

T1 - Can we accurately classify schizophrenia patients from healthy controls using magnetic resonance imaging and machine learning?

T2 - A multi-method and multi-dataset study

AU - Winterburn, Julie L

AU - Voineskos, Aristotle N

AU - Devenyi, Gabriel A

AU - Plitman, Eric

AU - de la Fuente-Sandoval, Camilo

AU - Bhagwat, Nikhil

AU - Graff-Guerrero, Ariel

AU - Knight, Jo

AU - Chakravarty, M Mallar

N1 - This is the author’s version of a work that was accepted for publication in Schizophrenia Research. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Schizophrenia Research 214, 2018 DOI: 10.1016/j.schres.2017.11.038

PY - 2019/12/1

Y1 - 2019/12/1

N2 - Machine learning is a powerful tool that has previously been used to classify schizophrenia (SZ) patients from healthy controls (HC) using magnetic resonance images. Each study, however, uses different datasets, classification algorithms, and validation techniques. Here, we perform a critical appraisal of the accuracy of machine learning methodologies used in SZ/HC classifications studies by comparing three machine learning algorithms (logistic regression [LR], support vector machines [SVMs], and linear discriminant analysis [LDA]) on three independent datasets (435 subjects total) using two tissue density estimates and cortical thickness (CT). Performance is assessed using 10-fold cross-validation, as well as a held-out validation set. Classification using CT outperformed tissue densities, but there was no clear effect of dataset. LR, SVMs, and LDA each yielded the highest accuracies for a different feature set and validation paradigm, but most accuracies were between 55 and 70%, well below previously reported values. The highest accuracy achieved was 73.5% using CT data and an SVM. Taken together, these results illustrate some of the obstacles to constructing effective disease classifiers, and suggest that tissue densities and CT may not be sufficiently sensitive for SZ/HC classification given current available methodologies and sample sizes.

AB - Machine learning is a powerful tool that has previously been used to classify schizophrenia (SZ) patients from healthy controls (HC) using magnetic resonance images. Each study, however, uses different datasets, classification algorithms, and validation techniques. Here, we perform a critical appraisal of the accuracy of machine learning methodologies used in SZ/HC classifications studies by comparing three machine learning algorithms (logistic regression [LR], support vector machines [SVMs], and linear discriminant analysis [LDA]) on three independent datasets (435 subjects total) using two tissue density estimates and cortical thickness (CT). Performance is assessed using 10-fold cross-validation, as well as a held-out validation set. Classification using CT outperformed tissue densities, but there was no clear effect of dataset. LR, SVMs, and LDA each yielded the highest accuracies for a different feature set and validation paradigm, but most accuracies were between 55 and 70%, well below previously reported values. The highest accuracy achieved was 73.5% using CT data and an SVM. Taken together, these results illustrate some of the obstacles to constructing effective disease classifiers, and suggest that tissue densities and CT may not be sufficiently sensitive for SZ/HC classification given current available methodologies and sample sizes.

KW - Structural magnetic resonance imaging

KW - Machine learning

KW - Classification

KW - Schizophrenia

KW - Voxel-based morphometry

KW - Cortical thickness

U2 - 10.1016/j.schres.2017.11.038

DO - 10.1016/j.schres.2017.11.038

M3 - Journal article

C2 - 29274736

VL - 214

SP - 3

EP - 10

JO - Schizophrenia Research

JF - Schizophrenia Research

SN - 0920-9964

ER -