Monte-Carlo based neuroimaging set-level multiple-comparison correction

Kewei Chen; Eric M. Reiman; Daniel Bandy; Gene E. Alexander

doi:10.1016/S1474-6670(17)33464-X

Monte-Carlo based neuroimaging set-level multiple-comparison correction

Kewei Chen
, Eric M. Reiman
, Daniel Bandy
, Gene E. Alexander

Research output: Contribution to journal › Conference article › peer-review

5 Scopus citations

Abstract

In neuroimaging studies, the type 1 error attributable to the number of independent regional comparisons should be considered. Implemented algorithms approximate the type 1 errors assuming Gaussian or Gaussian-based distribution, are typically for a contiguous/convex volume, and are for conventional hypothesis related to at least one or more clusters of certain size/height. As a complement, a Monte-Carlo algorithm is proposed to calculate the type 1 error of various types of hypotheses a researcher might have, and is applicable to Gaussian and non-Gaussian distributions over non-contiguous, non-convex and convoluted search volume.

Original language	English (US)
Pages (from-to)	11-15
Number of pages	5
Journal	IFAC Proceedings Volumes (IFAC-PapersOnline)
Volume	36
Issue number	15
DOIs	https://doi.org/10.1016/S1474-6670(17)33464-X
State	Published - 2003
Event	5th IFAC Symposium on Modelling and Control in Biomedical Systems 2003 - Melbourne, Australia Duration: Aug 21 2003 → Aug 23 2003

Keywords

Image analysis
Image processing
Magnetic resonance imaging
Monte-carlo simulation
Positron emission tomography
Statistical inference

ASJC Scopus subject areas

Control and Systems Engineering

Access to Document

10.1016/S1474-6670(17)33464-X

Cite this

@article{9f0bbb0c11394eab8f57375b15f23509,

title = "Monte-Carlo based neuroimaging set-level multiple-comparison correction",

abstract = "In neuroimaging studies, the type 1 error attributable to the number of independent regional comparisons should be considered. Implemented algorithms approximate the type 1 errors assuming Gaussian or Gaussian-based distribution, are typically for a contiguous/convex volume, and are for conventional hypothesis related to at least one or more clusters of certain size/height. As a complement, a Monte-Carlo algorithm is proposed to calculate the type 1 error of various types of hypotheses a researcher might have, and is applicable to Gaussian and non-Gaussian distributions over non-contiguous, non-convex and convoluted search volume.",

keywords = "Image analysis, Image processing, Magnetic resonance imaging, Monte-carlo simulation, Positron emission tomography, Statistical inference",

author = "Kewei Chen and Reiman, \{Eric M.\} and Daniel Bandy and Alexander, \{Gene E.\}",

note = "Publisher Copyright: Copyright {\textcopyright} 2003 IF AC.; 5th IFAC Symposium on Modelling and Control in Biomedical Systems 2003 ; Conference date: 21-08-2003 Through 23-08-2003",

year = "2003",

doi = "10.1016/S1474-6670(17)33464-X",

language = "English (US)",

volume = "36",

pages = "11--15",

journal = "IFAC Proceedings Volumes (IFAC-PapersOnline)",

issn = "1474-6670",

publisher = "IFAC Secretariat",

number = "15",

}

TY - JOUR

T1 - Monte-Carlo based neuroimaging set-level multiple-comparison correction

AU - Chen, Kewei

AU - Reiman, Eric M.

AU - Bandy, Daniel

AU - Alexander, Gene E.

PY - 2003

Y1 - 2003

N2 - In neuroimaging studies, the type 1 error attributable to the number of independent regional comparisons should be considered. Implemented algorithms approximate the type 1 errors assuming Gaussian or Gaussian-based distribution, are typically for a contiguous/convex volume, and are for conventional hypothesis related to at least one or more clusters of certain size/height. As a complement, a Monte-Carlo algorithm is proposed to calculate the type 1 error of various types of hypotheses a researcher might have, and is applicable to Gaussian and non-Gaussian distributions over non-contiguous, non-convex and convoluted search volume.

AB - In neuroimaging studies, the type 1 error attributable to the number of independent regional comparisons should be considered. Implemented algorithms approximate the type 1 errors assuming Gaussian or Gaussian-based distribution, are typically for a contiguous/convex volume, and are for conventional hypothesis related to at least one or more clusters of certain size/height. As a complement, a Monte-Carlo algorithm is proposed to calculate the type 1 error of various types of hypotheses a researcher might have, and is applicable to Gaussian and non-Gaussian distributions over non-contiguous, non-convex and convoluted search volume.

KW - Image analysis

KW - Image processing

KW - Magnetic resonance imaging

KW - Monte-carlo simulation

KW - Positron emission tomography

KW - Statistical inference

UR - https://www.scopus.com/pages/publications/48149106488

UR - https://www.scopus.com/pages/publications/48149106488#tab=citedBy

U2 - 10.1016/S1474-6670(17)33464-X

DO - 10.1016/S1474-6670(17)33464-X

M3 - Conference article

AN - SCOPUS:48149106488

SN - 1474-6670

VL - 36

SP - 11

EP - 15

JO - IFAC Proceedings Volumes (IFAC-PapersOnline)

JF - IFAC Proceedings Volumes (IFAC-PapersOnline)

IS - 15

T2 - 5th IFAC Symposium on Modelling and Control in Biomedical Systems 2003

Y2 - 21 August 2003 through 23 August 2003

ER -

Monte-Carlo based neuroimaging set-level multiple-comparison correction

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this