A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach

M. F. Ochs, Radka Stoyanova, F. Arias-Mendoza, T. R. Brown

Research output: Contribution to journalArticle

56 Citations (Scopus)

Abstract

A frequent problem in analysis is the need to find two matrices, closely related to the underlying measurement process, which when multiplied together reproduce the matrix of data points. Such problems arise throughout science, for example, in imaging where both the calibration of the sensor and the true scene may be unknown and in localized spectroscopy where multiple components may be present in varying amounts in any spectrum. Since both matrices are unknown, such a decomposition is a bilinear problem. We report here a solution to this problem for the case in which the decomposition results in matrices with elements drawn from positive additive distributions. We demonstrate the power of the methodology on chemical shift images (CSI). The new method, Bayesian spectral decomposition (BSD), reduces the CSI data to a small number of basis spectra together with their localized amplitudes. We apply this new algorithm to a 19F nonlocalized study of the catabolism of 5-fluorouracil in human liver, 31P CSI studies of a human head and calf muscle, and simulations which show its strengths and limitations. In all cases, the dataset, viewed as a matrix with rows containing the individual NMR spectra, results from the multiplication of a matrix of generally nonorthogonal basis spectra (the spectral matrix) by a matrix of the amplitudes of each basis spectrum in the the individual voxels (the amplitude matrix). The results show that BSD can simultaneously determine both the basis spectra and their distribution. In principle, BSD should solve this bilinear problem for any dataset which results from multiplication of matrices representing positive additive distributions if the data overdetermine the solutions.

Original languageEnglish
Pages (from-to)161-176
Number of pages16
JournalJournal of Magnetic Resonance
Volume137
Issue number1
StatePublished - Mar 1 1999
Externally publishedYes

Fingerprint

Bayes Theorem
Decomposition
decomposition
matrices
Fluorouracil
Calibration
Spectrum Analysis
Head
Chemical shift
Muscles
Liver
chemical equilibrium
multiplication
catabolism
Datasets
calves
muscles
liver
Muscle
Nuclear magnetic resonance

Keywords

  • Bayesian methods
  • Bilinear forms
  • Mixture analysis
  • NMR spectroscopy
  • Positive additive distributions
  • Spectral analysis

ASJC Scopus subject areas

  • Molecular Biology
  • Physical and Theoretical Chemistry
  • Spectroscopy
  • Radiology Nuclear Medicine and imaging
  • Condensed Matter Physics

Cite this

A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach. / Ochs, M. F.; Stoyanova, Radka; Arias-Mendoza, F.; Brown, T. R.

In: Journal of Magnetic Resonance, Vol. 137, No. 1, 01.03.1999, p. 161-176.

Research output: Contribution to journalArticle

Ochs, MF, Stoyanova, R, Arias-Mendoza, F & Brown, TR 1999, 'A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach', Journal of Magnetic Resonance, vol. 137, no. 1, pp. 161-176.
Ochs, M. F. ; Stoyanova, Radka ; Arias-Mendoza, F. ; Brown, T. R. / A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach. In: Journal of Magnetic Resonance. 1999 ; Vol. 137, No. 1. pp. 161-176.
@article{7be7c954372e45adaf6d6f79d5519e20,
title = "A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach",
abstract = "A frequent problem in analysis is the need to find two matrices, closely related to the underlying measurement process, which when multiplied together reproduce the matrix of data points. Such problems arise throughout science, for example, in imaging where both the calibration of the sensor and the true scene may be unknown and in localized spectroscopy where multiple components may be present in varying amounts in any spectrum. Since both matrices are unknown, such a decomposition is a bilinear problem. We report here a solution to this problem for the case in which the decomposition results in matrices with elements drawn from positive additive distributions. We demonstrate the power of the methodology on chemical shift images (CSI). The new method, Bayesian spectral decomposition (BSD), reduces the CSI data to a small number of basis spectra together with their localized amplitudes. We apply this new algorithm to a 19F nonlocalized study of the catabolism of 5-fluorouracil in human liver, 31P CSI studies of a human head and calf muscle, and simulations which show its strengths and limitations. In all cases, the dataset, viewed as a matrix with rows containing the individual NMR spectra, results from the multiplication of a matrix of generally nonorthogonal basis spectra (the spectral matrix) by a matrix of the amplitudes of each basis spectrum in the the individual voxels (the amplitude matrix). The results show that BSD can simultaneously determine both the basis spectra and their distribution. In principle, BSD should solve this bilinear problem for any dataset which results from multiplication of matrices representing positive additive distributions if the data overdetermine the solutions.",
keywords = "Bayesian methods, Bilinear forms, Mixture analysis, NMR spectroscopy, Positive additive distributions, Spectral analysis",
author = "Ochs, {M. F.} and Radka Stoyanova and F. Arias-Mendoza and Brown, {T. R.}",
year = "1999",
month = "3",
day = "1",
language = "English",
volume = "137",
pages = "161--176",
journal = "Journal of Magnetic Resonance",
issn = "1090-7807",
publisher = "Academic Press Inc.",
number = "1",

}

TY - JOUR

T1 - A New Method for Spectral Decomposition Using a Bilinear Bayesian Approach

AU - Ochs, M. F.

AU - Stoyanova, Radka

AU - Arias-Mendoza, F.

AU - Brown, T. R.

PY - 1999/3/1

Y1 - 1999/3/1

N2 - A frequent problem in analysis is the need to find two matrices, closely related to the underlying measurement process, which when multiplied together reproduce the matrix of data points. Such problems arise throughout science, for example, in imaging where both the calibration of the sensor and the true scene may be unknown and in localized spectroscopy where multiple components may be present in varying amounts in any spectrum. Since both matrices are unknown, such a decomposition is a bilinear problem. We report here a solution to this problem for the case in which the decomposition results in matrices with elements drawn from positive additive distributions. We demonstrate the power of the methodology on chemical shift images (CSI). The new method, Bayesian spectral decomposition (BSD), reduces the CSI data to a small number of basis spectra together with their localized amplitudes. We apply this new algorithm to a 19F nonlocalized study of the catabolism of 5-fluorouracil in human liver, 31P CSI studies of a human head and calf muscle, and simulations which show its strengths and limitations. In all cases, the dataset, viewed as a matrix with rows containing the individual NMR spectra, results from the multiplication of a matrix of generally nonorthogonal basis spectra (the spectral matrix) by a matrix of the amplitudes of each basis spectrum in the the individual voxels (the amplitude matrix). The results show that BSD can simultaneously determine both the basis spectra and their distribution. In principle, BSD should solve this bilinear problem for any dataset which results from multiplication of matrices representing positive additive distributions if the data overdetermine the solutions.

AB - A frequent problem in analysis is the need to find two matrices, closely related to the underlying measurement process, which when multiplied together reproduce the matrix of data points. Such problems arise throughout science, for example, in imaging where both the calibration of the sensor and the true scene may be unknown and in localized spectroscopy where multiple components may be present in varying amounts in any spectrum. Since both matrices are unknown, such a decomposition is a bilinear problem. We report here a solution to this problem for the case in which the decomposition results in matrices with elements drawn from positive additive distributions. We demonstrate the power of the methodology on chemical shift images (CSI). The new method, Bayesian spectral decomposition (BSD), reduces the CSI data to a small number of basis spectra together with their localized amplitudes. We apply this new algorithm to a 19F nonlocalized study of the catabolism of 5-fluorouracil in human liver, 31P CSI studies of a human head and calf muscle, and simulations which show its strengths and limitations. In all cases, the dataset, viewed as a matrix with rows containing the individual NMR spectra, results from the multiplication of a matrix of generally nonorthogonal basis spectra (the spectral matrix) by a matrix of the amplitudes of each basis spectrum in the the individual voxels (the amplitude matrix). The results show that BSD can simultaneously determine both the basis spectra and their distribution. In principle, BSD should solve this bilinear problem for any dataset which results from multiplication of matrices representing positive additive distributions if the data overdetermine the solutions.

KW - Bayesian methods

KW - Bilinear forms

KW - Mixture analysis

KW - NMR spectroscopy

KW - Positive additive distributions

KW - Spectral analysis

UR - http://www.scopus.com/inward/record.url?scp=0033093826&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033093826&partnerID=8YFLogxK

M3 - Article

C2 - 10053145

AN - SCOPUS:0033093826

VL - 137

SP - 161

EP - 176

JO - Journal of Magnetic Resonance

JF - Journal of Magnetic Resonance

SN - 1090-7807

IS - 1

ER -