Combining energy and Laplacian regularization to accurately retrieve the depth of brain activity of diffuse optical tomographic data

Antonio M. Chiarelli, Edward L. Maclin, Kathy A. Low, Kyle E. Mathewson, Monica Fabiani, Gabriele Gratton

Research output: Contribution to journalArticle

Abstract

Diffuse optical tomography (DOT) provides data about brain function using surface recordings. Despite recent advancements, an unbiased method for estimating the depth of absorption changes and for providing an accurate three-dimensional (3-D) reconstruction remains elusive. DOT involves solving an ill-posed inverse problem, requiring additional criteria for finding unique solutions. The most commonly used criterion is energy minimization (energy constraint). However, as measurements are taken from only one side of the medium (the scalp) and sensitivity is greater at shallow depths, the energy constraint leads to solutions that tend to be small and superficial. To correct for this bias, we combine the energy constraint with another criterion, minimization of spatial derivatives (Laplacian constraint, also used in low resolution electromagnetic tomography, LORETA). Used in isolation, the Laplacian constraint leads to solutions that tend to be large and deep. Using simulated, phantom, and actual brain activation data, we show that combining these two criteria results in accurate (error <2 mm) absorption depth estimates, while maintaining a two-point spatial resolution of <24 mm up to a depth of 30 mm. This indicates that accurate 3-D reconstruction of brain activity up to 30 mm from the scalp can be obtained with DOT.

Original languageEnglish (US)
Article number036008
JournalJournal of biomedical optics
Volume21
Issue number3
DOIs
StatePublished - Mar 1 2016

Fingerprint

Optical tomography
brain
Brain
tomography
Inverse problems
Tomography
energy
Chemical activation
optimization
Derivatives
isolation
estimating
spatial resolution
recording
activation
electromagnetism
sensitivity
estimates

Keywords

  • 3-D reconstruction
  • Laplacian regularization, low resolution electromagnetic tomography
  • diffuse optical tomography
  • functional brain mapping
  • minimum-norm
  • near infrared spectroscopy

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Atomic and Molecular Physics, and Optics
  • Biomedical Engineering

Cite this

Combining energy and Laplacian regularization to accurately retrieve the depth of brain activity of diffuse optical tomographic data. / Chiarelli, Antonio M.; Maclin, Edward L.; Low, Kathy A.; Mathewson, Kyle E.; Fabiani, Monica; Gratton, Gabriele.

In: Journal of biomedical optics, Vol. 21, No. 3, 036008, 01.03.2016.

Research output: Contribution to journalArticle

@article{21153de0ecc244d1bbe66b2c7ac91e77,
title = "Combining energy and Laplacian regularization to accurately retrieve the depth of brain activity of diffuse optical tomographic data",
abstract = "Diffuse optical tomography (DOT) provides data about brain function using surface recordings. Despite recent advancements, an unbiased method for estimating the depth of absorption changes and for providing an accurate three-dimensional (3-D) reconstruction remains elusive. DOT involves solving an ill-posed inverse problem, requiring additional criteria for finding unique solutions. The most commonly used criterion is energy minimization (energy constraint). However, as measurements are taken from only one side of the medium (the scalp) and sensitivity is greater at shallow depths, the energy constraint leads to solutions that tend to be small and superficial. To correct for this bias, we combine the energy constraint with another criterion, minimization of spatial derivatives (Laplacian constraint, also used in low resolution electromagnetic tomography, LORETA). Used in isolation, the Laplacian constraint leads to solutions that tend to be large and deep. Using simulated, phantom, and actual brain activation data, we show that combining these two criteria results in accurate (error <2 mm) absorption depth estimates, while maintaining a two-point spatial resolution of <24 mm up to a depth of 30 mm. This indicates that accurate 3-D reconstruction of brain activity up to 30 mm from the scalp can be obtained with DOT.",
keywords = "3-D reconstruction, Laplacian regularization, low resolution electromagnetic tomography, diffuse optical tomography, functional brain mapping, minimum-norm, near infrared spectroscopy",
author = "Chiarelli, {Antonio M.} and Maclin, {Edward L.} and Low, {Kathy A.} and Mathewson, {Kyle E.} and Monica Fabiani and Gabriele Gratton",
year = "2016",
month = "3",
day = "1",
doi = "10.1117/1.JBO.21.3.036008",
language = "English (US)",
volume = "21",
journal = "Journal of Biomedical Optics",
issn = "1083-3668",
publisher = "SPIE",
number = "3",

}

TY - JOUR

T1 - Combining energy and Laplacian regularization to accurately retrieve the depth of brain activity of diffuse optical tomographic data

AU - Chiarelli, Antonio M.

AU - Maclin, Edward L.

AU - Low, Kathy A.

AU - Mathewson, Kyle E.

AU - Fabiani, Monica

AU - Gratton, Gabriele

PY - 2016/3/1

Y1 - 2016/3/1

N2 - Diffuse optical tomography (DOT) provides data about brain function using surface recordings. Despite recent advancements, an unbiased method for estimating the depth of absorption changes and for providing an accurate three-dimensional (3-D) reconstruction remains elusive. DOT involves solving an ill-posed inverse problem, requiring additional criteria for finding unique solutions. The most commonly used criterion is energy minimization (energy constraint). However, as measurements are taken from only one side of the medium (the scalp) and sensitivity is greater at shallow depths, the energy constraint leads to solutions that tend to be small and superficial. To correct for this bias, we combine the energy constraint with another criterion, minimization of spatial derivatives (Laplacian constraint, also used in low resolution electromagnetic tomography, LORETA). Used in isolation, the Laplacian constraint leads to solutions that tend to be large and deep. Using simulated, phantom, and actual brain activation data, we show that combining these two criteria results in accurate (error <2 mm) absorption depth estimates, while maintaining a two-point spatial resolution of <24 mm up to a depth of 30 mm. This indicates that accurate 3-D reconstruction of brain activity up to 30 mm from the scalp can be obtained with DOT.

AB - Diffuse optical tomography (DOT) provides data about brain function using surface recordings. Despite recent advancements, an unbiased method for estimating the depth of absorption changes and for providing an accurate three-dimensional (3-D) reconstruction remains elusive. DOT involves solving an ill-posed inverse problem, requiring additional criteria for finding unique solutions. The most commonly used criterion is energy minimization (energy constraint). However, as measurements are taken from only one side of the medium (the scalp) and sensitivity is greater at shallow depths, the energy constraint leads to solutions that tend to be small and superficial. To correct for this bias, we combine the energy constraint with another criterion, minimization of spatial derivatives (Laplacian constraint, also used in low resolution electromagnetic tomography, LORETA). Used in isolation, the Laplacian constraint leads to solutions that tend to be large and deep. Using simulated, phantom, and actual brain activation data, we show that combining these two criteria results in accurate (error <2 mm) absorption depth estimates, while maintaining a two-point spatial resolution of <24 mm up to a depth of 30 mm. This indicates that accurate 3-D reconstruction of brain activity up to 30 mm from the scalp can be obtained with DOT.

KW - 3-D reconstruction

KW - Laplacian regularization, low resolution electromagnetic tomography

KW - diffuse optical tomography

KW - functional brain mapping

KW - minimum-norm

KW - near infrared spectroscopy

UR - http://www.scopus.com/inward/record.url?scp=84964505039&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964505039&partnerID=8YFLogxK

U2 - 10.1117/1.JBO.21.3.036008

DO - 10.1117/1.JBO.21.3.036008

M3 - Article

C2 - 26987429

AN - SCOPUS:84964505039

VL - 21

JO - Journal of Biomedical Optics

JF - Journal of Biomedical Optics

SN - 1083-3668

IS - 3

M1 - 036008

ER -