ML segmentation strategies for object interference compensation in FDG-PET lesion quantification

E. de Bernardi, F. Fiorani Gallotta, C. Gianoli, F. Zito, P. Gerundini, G. Baselli

Research output: Contribution to journalArticle

Abstract

Background: Quantification of lesion activity by FDG uptake in oncological PET is severely limited by partial volume effects. A maximum likelihood (ML) expectation maximization (EM) algorithm considering regional basis functions (AWOSEM-region) had been previously developed. Regional basis functions are iteratively segmented and quantified, thus identifying the volume and the activity of the lesion. Objectives: Improvement of AWOSEM-region when analyzing proximal interfering hot objects is addressed by proper segmentation initialization steps and models of spill-out and partial volume effects. Conditions relevant to lung PET-CT studies are considered: 1) lesion close to hot organ (e.g. chest wall, heart and mediastinum), 2) two close lesions. Methods: CT image was considered for presegmenting hot anatomical structures, never for lesion identification, solely defined by iterations on PET data. Further resolution recovery beyond the smooth standard clinical image was necessary to start lesion segmentation. A watershed algorithm was used to separate two close lesions. A subtraction of the spill-out from a nearby hot organ was introduced to enhance a lesion for the initial segmentation and start the further quantification steps. Biograph scanner blurring was modeled from phantom data in order to implement the procedure for 3D clinical lung studies. Results: In simulations, the procedure was able to separate structures as close as one pixel-size (2.25 mm). Robustness against the input segmentation errors defining the addressed objects was tested showing that convergence was not sensitive to initial volume overestimates up to 130%. Poor robustness was found against underestimates. A clinical study of a small lung lesion close to chest wall displayed a good recovery of both lesion activity and volume. Conclusions: With proper initialization and models of spill-out from hot organs, AWOSEMregion can be successfully applied to lung oncological studies.

Original languageEnglish
Pages (from-to)537-541
Number of pages5
JournalMethods of Information in Medicine
Volume49
Issue number5
DOIs
Publication statusPublished - 2010

Fingerprint

Lung
Thoracic Wall
Mediastinum
Clinical Studies
Positron Emission Tomography Computed Tomography

Keywords

  • AWOSEM
  • Lesion quantification
  • Partial volume effect correction
  • PET-CT
  • Targeted reconstruction

ASJC Scopus subject areas

  • Health Informatics
  • Advanced and Specialised Nursing
  • Health Information Management

Cite this

ML segmentation strategies for object interference compensation in FDG-PET lesion quantification. / de Bernardi, E.; Fiorani Gallotta, F.; Gianoli, C.; Zito, F.; Gerundini, P.; Baselli, G.

In: Methods of Information in Medicine, Vol. 49, No. 5, 2010, p. 537-541.

Research output: Contribution to journalArticle

de Bernardi, E. ; Fiorani Gallotta, F. ; Gianoli, C. ; Zito, F. ; Gerundini, P. ; Baselli, G. / ML segmentation strategies for object interference compensation in FDG-PET lesion quantification. In: Methods of Information in Medicine. 2010 ; Vol. 49, No. 5. pp. 537-541.
@article{844f1e1a18f546d7bf17a58a84983c06,
title = "ML segmentation strategies for object interference compensation in FDG-PET lesion quantification",
abstract = "Background: Quantification of lesion activity by FDG uptake in oncological PET is severely limited by partial volume effects. A maximum likelihood (ML) expectation maximization (EM) algorithm considering regional basis functions (AWOSEM-region) had been previously developed. Regional basis functions are iteratively segmented and quantified, thus identifying the volume and the activity of the lesion. Objectives: Improvement of AWOSEM-region when analyzing proximal interfering hot objects is addressed by proper segmentation initialization steps and models of spill-out and partial volume effects. Conditions relevant to lung PET-CT studies are considered: 1) lesion close to hot organ (e.g. chest wall, heart and mediastinum), 2) two close lesions. Methods: CT image was considered for presegmenting hot anatomical structures, never for lesion identification, solely defined by iterations on PET data. Further resolution recovery beyond the smooth standard clinical image was necessary to start lesion segmentation. A watershed algorithm was used to separate two close lesions. A subtraction of the spill-out from a nearby hot organ was introduced to enhance a lesion for the initial segmentation and start the further quantification steps. Biograph scanner blurring was modeled from phantom data in order to implement the procedure for 3D clinical lung studies. Results: In simulations, the procedure was able to separate structures as close as one pixel-size (2.25 mm). Robustness against the input segmentation errors defining the addressed objects was tested showing that convergence was not sensitive to initial volume overestimates up to 130{\%}. Poor robustness was found against underestimates. A clinical study of a small lung lesion close to chest wall displayed a good recovery of both lesion activity and volume. Conclusions: With proper initialization and models of spill-out from hot organs, AWOSEMregion can be successfully applied to lung oncological studies.",
keywords = "AWOSEM, Lesion quantification, Partial volume effect correction, PET-CT, Targeted reconstruction",
author = "{de Bernardi}, E. and {Fiorani Gallotta}, F. and C. Gianoli and F. Zito and P. Gerundini and G. Baselli",
year = "2010",
doi = "10.3414/ME09-02-0040",
language = "English",
volume = "49",
pages = "537--541",
journal = "Methods of Information in Medicine",
issn = "0026-1270",
publisher = "Schattauer GmbH",
number = "5",

}

TY - JOUR

T1 - ML segmentation strategies for object interference compensation in FDG-PET lesion quantification

AU - de Bernardi, E.

AU - Fiorani Gallotta, F.

AU - Gianoli, C.

AU - Zito, F.

AU - Gerundini, P.

AU - Baselli, G.

PY - 2010

Y1 - 2010

N2 - Background: Quantification of lesion activity by FDG uptake in oncological PET is severely limited by partial volume effects. A maximum likelihood (ML) expectation maximization (EM) algorithm considering regional basis functions (AWOSEM-region) had been previously developed. Regional basis functions are iteratively segmented and quantified, thus identifying the volume and the activity of the lesion. Objectives: Improvement of AWOSEM-region when analyzing proximal interfering hot objects is addressed by proper segmentation initialization steps and models of spill-out and partial volume effects. Conditions relevant to lung PET-CT studies are considered: 1) lesion close to hot organ (e.g. chest wall, heart and mediastinum), 2) two close lesions. Methods: CT image was considered for presegmenting hot anatomical structures, never for lesion identification, solely defined by iterations on PET data. Further resolution recovery beyond the smooth standard clinical image was necessary to start lesion segmentation. A watershed algorithm was used to separate two close lesions. A subtraction of the spill-out from a nearby hot organ was introduced to enhance a lesion for the initial segmentation and start the further quantification steps. Biograph scanner blurring was modeled from phantom data in order to implement the procedure for 3D clinical lung studies. Results: In simulations, the procedure was able to separate structures as close as one pixel-size (2.25 mm). Robustness against the input segmentation errors defining the addressed objects was tested showing that convergence was not sensitive to initial volume overestimates up to 130%. Poor robustness was found against underestimates. A clinical study of a small lung lesion close to chest wall displayed a good recovery of both lesion activity and volume. Conclusions: With proper initialization and models of spill-out from hot organs, AWOSEMregion can be successfully applied to lung oncological studies.

AB - Background: Quantification of lesion activity by FDG uptake in oncological PET is severely limited by partial volume effects. A maximum likelihood (ML) expectation maximization (EM) algorithm considering regional basis functions (AWOSEM-region) had been previously developed. Regional basis functions are iteratively segmented and quantified, thus identifying the volume and the activity of the lesion. Objectives: Improvement of AWOSEM-region when analyzing proximal interfering hot objects is addressed by proper segmentation initialization steps and models of spill-out and partial volume effects. Conditions relevant to lung PET-CT studies are considered: 1) lesion close to hot organ (e.g. chest wall, heart and mediastinum), 2) two close lesions. Methods: CT image was considered for presegmenting hot anatomical structures, never for lesion identification, solely defined by iterations on PET data. Further resolution recovery beyond the smooth standard clinical image was necessary to start lesion segmentation. A watershed algorithm was used to separate two close lesions. A subtraction of the spill-out from a nearby hot organ was introduced to enhance a lesion for the initial segmentation and start the further quantification steps. Biograph scanner blurring was modeled from phantom data in order to implement the procedure for 3D clinical lung studies. Results: In simulations, the procedure was able to separate structures as close as one pixel-size (2.25 mm). Robustness against the input segmentation errors defining the addressed objects was tested showing that convergence was not sensitive to initial volume overestimates up to 130%. Poor robustness was found against underestimates. A clinical study of a small lung lesion close to chest wall displayed a good recovery of both lesion activity and volume. Conclusions: With proper initialization and models of spill-out from hot organs, AWOSEMregion can be successfully applied to lung oncological studies.

KW - AWOSEM

KW - Lesion quantification

KW - Partial volume effect correction

KW - PET-CT

KW - Targeted reconstruction

UR - http://www.scopus.com/inward/record.url?scp=79952277255&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952277255&partnerID=8YFLogxK

U2 - 10.3414/ME09-02-0040

DO - 10.3414/ME09-02-0040

M3 - Article

C2 - 20490426

AN - SCOPUS:79952277255

VL - 49

SP - 537

EP - 541

JO - Methods of Information in Medicine

JF - Methods of Information in Medicine

SN - 0026-1270

IS - 5

ER -