Calculation of intravascular signal in dynamic contrast enhanced-MRI using adaptive complex independent component analysis

Hatef Mehrabian, Rajiv Chopra, Anne L. Martel

Research output: Contribution to journalArticle

22 Citations (Scopus)

Abstract

Assessing tumor response to therapy is a crucial step in personalized treatments. Pharmacokinetic (PK) modeling provides quantitative information about tumor perfusion and vascular permeability that are associated with prognostic factors. A fundamental step in most PK analyses is calculating the signal that is generated in the tumor vasculature. This signal is usually inseparable from the extravascular extracellular signal. It was shown previously using in vivo and phantom experiments that independent component analysis (ICA) is capable of calculating the intravascular time-intensity curve in dynamic contrast enhanced (DCE)-MRI. A novel adaptive complex independent component analysis (AC-ICA) technique is developed in this study to calculate the intravascular time-intensity curve and separate this signal from the DCE-MR images of tumors. The use of the complex-valued DCE-MRI images rather than the commonly used magnitude images satisfied the fundamental assumption of ICA, i.e., linear mixing of the sources. Using an adaptive cost function in ICA through estimating the probability distribution of the tumor vasculature at each iteration resulted in a more robust and accurate separation algorithm. The AC-ICA algorithm provided a better estimate for the intravascular time-intensity curve than the previous ICA-based method.

Original languageEnglish (US)
Article number6380622
Pages (from-to)699-710
Number of pages12
JournalIEEE Transactions on Medical Imaging
Volume32
Issue number4
DOIs
StatePublished - 2013

Fingerprint

Independent component analysis
Magnetic resonance imaging
Tumors
Neoplasms
Pharmacokinetics
Capillary Permeability
Cost functions
Probability distributions
Perfusion
Costs and Cost Analysis
Experiments

Keywords

  • Adaptive complex independent component analysis (AC-ICA)
  • arterial input function (AIF)
  • intravascular signal intensity
  • pharmacokinetic modeling

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Science Applications
  • Radiological and Ultrasound Technology
  • Software

Cite this

Calculation of intravascular signal in dynamic contrast enhanced-MRI using adaptive complex independent component analysis. / Mehrabian, Hatef; Chopra, Rajiv; Martel, Anne L.

In: IEEE Transactions on Medical Imaging, Vol. 32, No. 4, 6380622, 2013, p. 699-710.

Research output: Contribution to journalArticle

@article{2727dd97d0c242a09fb57a922675a182,
title = "Calculation of intravascular signal in dynamic contrast enhanced-MRI using adaptive complex independent component analysis",
abstract = "Assessing tumor response to therapy is a crucial step in personalized treatments. Pharmacokinetic (PK) modeling provides quantitative information about tumor perfusion and vascular permeability that are associated with prognostic factors. A fundamental step in most PK analyses is calculating the signal that is generated in the tumor vasculature. This signal is usually inseparable from the extravascular extracellular signal. It was shown previously using in vivo and phantom experiments that independent component analysis (ICA) is capable of calculating the intravascular time-intensity curve in dynamic contrast enhanced (DCE)-MRI. A novel adaptive complex independent component analysis (AC-ICA) technique is developed in this study to calculate the intravascular time-intensity curve and separate this signal from the DCE-MR images of tumors. The use of the complex-valued DCE-MRI images rather than the commonly used magnitude images satisfied the fundamental assumption of ICA, i.e., linear mixing of the sources. Using an adaptive cost function in ICA through estimating the probability distribution of the tumor vasculature at each iteration resulted in a more robust and accurate separation algorithm. The AC-ICA algorithm provided a better estimate for the intravascular time-intensity curve than the previous ICA-based method.",
keywords = "Adaptive complex independent component analysis (AC-ICA), arterial input function (AIF), intravascular signal intensity, pharmacokinetic modeling",
author = "Hatef Mehrabian and Rajiv Chopra and Martel, {Anne L.}",
year = "2013",
doi = "10.1109/TMI.2012.2233747",
language = "English (US)",
volume = "32",
pages = "699--710",
journal = "IEEE Transactions on Medical Imaging",
issn = "0278-0062",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "4",

}

TY - JOUR

T1 - Calculation of intravascular signal in dynamic contrast enhanced-MRI using adaptive complex independent component analysis

AU - Mehrabian, Hatef

AU - Chopra, Rajiv

AU - Martel, Anne L.

PY - 2013

Y1 - 2013

N2 - Assessing tumor response to therapy is a crucial step in personalized treatments. Pharmacokinetic (PK) modeling provides quantitative information about tumor perfusion and vascular permeability that are associated with prognostic factors. A fundamental step in most PK analyses is calculating the signal that is generated in the tumor vasculature. This signal is usually inseparable from the extravascular extracellular signal. It was shown previously using in vivo and phantom experiments that independent component analysis (ICA) is capable of calculating the intravascular time-intensity curve in dynamic contrast enhanced (DCE)-MRI. A novel adaptive complex independent component analysis (AC-ICA) technique is developed in this study to calculate the intravascular time-intensity curve and separate this signal from the DCE-MR images of tumors. The use of the complex-valued DCE-MRI images rather than the commonly used magnitude images satisfied the fundamental assumption of ICA, i.e., linear mixing of the sources. Using an adaptive cost function in ICA through estimating the probability distribution of the tumor vasculature at each iteration resulted in a more robust and accurate separation algorithm. The AC-ICA algorithm provided a better estimate for the intravascular time-intensity curve than the previous ICA-based method.

AB - Assessing tumor response to therapy is a crucial step in personalized treatments. Pharmacokinetic (PK) modeling provides quantitative information about tumor perfusion and vascular permeability that are associated with prognostic factors. A fundamental step in most PK analyses is calculating the signal that is generated in the tumor vasculature. This signal is usually inseparable from the extravascular extracellular signal. It was shown previously using in vivo and phantom experiments that independent component analysis (ICA) is capable of calculating the intravascular time-intensity curve in dynamic contrast enhanced (DCE)-MRI. A novel adaptive complex independent component analysis (AC-ICA) technique is developed in this study to calculate the intravascular time-intensity curve and separate this signal from the DCE-MR images of tumors. The use of the complex-valued DCE-MRI images rather than the commonly used magnitude images satisfied the fundamental assumption of ICA, i.e., linear mixing of the sources. Using an adaptive cost function in ICA through estimating the probability distribution of the tumor vasculature at each iteration resulted in a more robust and accurate separation algorithm. The AC-ICA algorithm provided a better estimate for the intravascular time-intensity curve than the previous ICA-based method.

KW - Adaptive complex independent component analysis (AC-ICA)

KW - arterial input function (AIF)

KW - intravascular signal intensity

KW - pharmacokinetic modeling

UR - http://www.scopus.com/inward/record.url?scp=84875983151&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875983151&partnerID=8YFLogxK

U2 - 10.1109/TMI.2012.2233747

DO - 10.1109/TMI.2012.2233747

M3 - Article

C2 - 23247848

AN - SCOPUS:84875983151

VL - 32

SP - 699

EP - 710

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

SN - 0278-0062

IS - 4

M1 - 6380622

ER -