Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging

ENIGMA

doi:10.1007/978-3-319-67389-9_43

Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging

ENIGMA

Research output: Contribution to journal › Article (Academic Journal) › peer-review

3 Citations (Scopus)

218 Downloads (Pure)

Abstract

As very large studies of complex neuroimaging phenotypes become more common, human quality assessment of MRI-derived data remains one of the last major bottlenecks. Few attempts have so far been made to address this issue with machine learning. In this work, we optimize predictive models of quality for meshes representing deep brain structure shapes. We use standard vertex-wise and global shape features computed homologously across 19 cohorts and over 7500 human-rated subjects, training kernelized Support Vector Machine and Gradient Boosted Decision Trees classifiers to detect meshes of failing quality. Our models generalize across datasets and diseases, reducing human workload by 30-70%, or equivalently hundreds of human rater hours for datasets of comparable size, with recall rates approaching inter-rater reliability.

Original language	English
Pages (from-to)	371-378
Number of pages	8
Journal	Machine learning in medical imaging. MLMI (Workshop)
Volume	10541
Early online date	7 Sept 2017
DOIs	https://doi.org/10.1007/978-3-319-67389-9_43
Publication status	E-pub ahead of print - 7 Sept 2017

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-319-67389-9_43

Full-text PDF (accepted author manuscript)
This is the accepted author manuscript (AAM). The final published version (version of record) is available online via Springer Link at https://doi.org/10.1007/978-3-319-67389-9_43 . Please refer to any applicable terms of use of the publisher.
Accepted author manuscript, 1.51 MBLicence: Other

Handle.net

Persistent link

Cite this

@article{8fe0fb18851a4e4f95367825a0f25ec7,

title = "Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging",

abstract = "As very large studies of complex neuroimaging phenotypes become more common, human quality assessment of MRI-derived data remains one of the last major bottlenecks. Few attempts have so far been made to address this issue with machine learning. In this work, we optimize predictive models of quality for meshes representing deep brain structure shapes. We use standard vertex-wise and global shape features computed homologously across 19 cohorts and over 7500 human-rated subjects, training kernelized Support Vector Machine and Gradient Boosted Decision Trees classifiers to detect meshes of failing quality. Our models generalize across datasets and diseases, reducing human workload by 30-70%, or equivalently hundreds of human rater hours for datasets of comparable size, with recall rates approaching inter-rater reliability.",

author = "ENIGMA and Dmitry Petrov and Gutman, {Boris A} and Yu, {Shih-Hua Julie} and {van Erp}, {Theo G M} and Lianne Schmaal and Dick Veltman and Kathryn Alpert and Dmitry Isaev and Artemis Zavaliangos-Petropulu and Ching, {Christopher R K} and Vince Calhoun and David Glahn and Satterthwaite, {Theodore D} and Andreasen, {Ole Andreas} and Stefan Borgwardt and Fleur Howells and Nynke Groenewold and Aristotle Voineskos and Joaquim Radua and Potkin, {Steven G} and Benedicto Crespo-Facorro and Diana Tordesillas-Guti{\'e}rrez and Li Shen and Irina Lebedeva and Gianfranco Spalletta and Gary Donohoe and Peter Kochunov and Rosa, {Pedro G P} and Anthony James and Udo Dannlowski and Baune, {Bernhard T} and Andr{\'e} Aleman and Gotlib, {Ian H} and Henrik Walter and Martin Walter and Soares, {Jair C} and Stefan Ehrlich and Gur, {Ruben C} and Doan, {N Trung} and Ingrid Agartz and Westlye, {Lars T} and Fabienne Harrisberger and Anita Riecher-R{\"o}ssler and Anne Uhlmann and Stein, {Dan J} and Dickie, {Erin W} and Edith Pomarol-Clotet and Paola Fuentes-Claramonte and Canales-Rodr{\'i}guez, {Erick Jorge} and Esther Walton",

year = "2017",

month = sep,

day = "7",

doi = "10.1007/978-3-319-67389-9_43",

language = "English",

volume = "10541",

pages = "371--378",

journal = "Machine learning in medical imaging. MLMI (Workshop)",

}

TY - JOUR

T1 - Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging

AU - ENIGMA

AU - Petrov, Dmitry

AU - Gutman, Boris A

AU - Yu, Shih-Hua Julie

AU - van Erp, Theo G M

AU - Schmaal, Lianne

AU - Veltman, Dick

AU - Alpert, Kathryn

AU - Isaev, Dmitry

AU - Zavaliangos-Petropulu, Artemis

AU - Ching, Christopher R K

AU - Calhoun, Vince

AU - Glahn, David

AU - Satterthwaite, Theodore D

AU - Andreasen, Ole Andreas

AU - Borgwardt, Stefan

AU - Howells, Fleur

AU - Groenewold, Nynke

AU - Voineskos, Aristotle

AU - Radua, Joaquim

AU - Potkin, Steven G

AU - Crespo-Facorro, Benedicto

AU - Tordesillas-Gutiérrez, Diana

AU - Shen, Li

AU - Lebedeva, Irina

AU - Spalletta, Gianfranco

AU - Donohoe, Gary

AU - Kochunov, Peter

AU - Rosa, Pedro G P

AU - James, Anthony

AU - Dannlowski, Udo

AU - Baune, Bernhard T

AU - Aleman, André

AU - Gotlib, Ian H

AU - Walter, Henrik

AU - Walter, Martin

AU - Soares, Jair C

AU - Ehrlich, Stefan

AU - Gur, Ruben C

AU - Doan, N Trung

AU - Agartz, Ingrid

AU - Westlye, Lars T

AU - Harrisberger, Fabienne

AU - Riecher-Rössler, Anita

AU - Uhlmann, Anne

AU - Stein, Dan J

AU - Dickie, Erin W

AU - Pomarol-Clotet, Edith

AU - Fuentes-Claramonte, Paola

AU - Canales-Rodríguez, Erick Jorge

AU - Walton, Esther

PY - 2017/9/7

Y1 - 2017/9/7

N2 - As very large studies of complex neuroimaging phenotypes become more common, human quality assessment of MRI-derived data remains one of the last major bottlenecks. Few attempts have so far been made to address this issue with machine learning. In this work, we optimize predictive models of quality for meshes representing deep brain structure shapes. We use standard vertex-wise and global shape features computed homologously across 19 cohorts and over 7500 human-rated subjects, training kernelized Support Vector Machine and Gradient Boosted Decision Trees classifiers to detect meshes of failing quality. Our models generalize across datasets and diseases, reducing human workload by 30-70%, or equivalently hundreds of human rater hours for datasets of comparable size, with recall rates approaching inter-rater reliability.

AB - As very large studies of complex neuroimaging phenotypes become more common, human quality assessment of MRI-derived data remains one of the last major bottlenecks. Few attempts have so far been made to address this issue with machine learning. In this work, we optimize predictive models of quality for meshes representing deep brain structure shapes. We use standard vertex-wise and global shape features computed homologously across 19 cohorts and over 7500 human-rated subjects, training kernelized Support Vector Machine and Gradient Boosted Decision Trees classifiers to detect meshes of failing quality. Our models generalize across datasets and diseases, reducing human workload by 30-70%, or equivalently hundreds of human rater hours for datasets of comparable size, with recall rates approaching inter-rater reliability.

U2 - 10.1007/978-3-319-67389-9_43

DO - 10.1007/978-3-319-67389-9_43

M3 - Article (Academic Journal)

C2 - 30035274

VL - 10541

SP - 371

EP - 378

JO - Machine learning in medical imaging. MLMI (Workshop)

JF - Machine learning in medical imaging. MLMI (Workshop)

ER -

Machine Learning for Large-Scale Quality Control of 3D Shape Models in Neuroimaging

Abstract

UN SDGs

Access to Document

Handle.net

Fingerprint

Cite this