A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images

Gang Lin, Monica K. Chawla, Kathy Olson, Carol A Barnes, John F. Guzowski, Christopher Bjornsson, William Shain, Badrinath Roysam

Research output: Contribution to journalArticle

85 Citations (Scopus)

Abstract

Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model-based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user-provided training examples. Starting with an initial over-segmentation produced by a gradient-weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7% and classification accuracy was 93.5%, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi-model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.

Original languageEnglish (US)
Pages (from-to)724-736
Number of pages13
JournalCytometry Part A
Volume71
Issue number9
DOIs
StatePublished - Sep 2007

Fingerprint

Cell Nucleus
Population
Discriminant Analysis
Brain

Keywords

  • 3D confocal microscopy
  • Batch processing
  • Bayesian estimator
  • Cell nuclei
  • Classification
  • Model-based
  • Parzen window
  • Region merging
  • Segmentation
  • Watershed algorithm

ASJC Scopus subject areas

  • Hematology
  • Cell Biology
  • Pathology and Forensic Medicine
  • Biophysics
  • Endocrinology

Cite this

A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images. / Lin, Gang; Chawla, Monica K.; Olson, Kathy; Barnes, Carol A; Guzowski, John F.; Bjornsson, Christopher; Shain, William; Roysam, Badrinath.

In: Cytometry Part A, Vol. 71, No. 9, 09.2007, p. 724-736.

Research output: Contribution to journalArticle

Lin, Gang ; Chawla, Monica K. ; Olson, Kathy ; Barnes, Carol A ; Guzowski, John F. ; Bjornsson, Christopher ; Shain, William ; Roysam, Badrinath. / A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images. In: Cytometry Part A. 2007 ; Vol. 71, No. 9. pp. 724-736.
@article{45233a25d48c47a0a14427e585cc43d9,
title = "A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images",
abstract = "Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model-based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user-provided training examples. Starting with an initial over-segmentation produced by a gradient-weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7{\%} and classification accuracy was 93.5{\%}, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi-model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.",
keywords = "3D confocal microscopy, Batch processing, Bayesian estimator, Cell nuclei, Classification, Model-based, Parzen window, Region merging, Segmentation, Watershed algorithm",
author = "Gang Lin and Chawla, {Monica K.} and Kathy Olson and Barnes, {Carol A} and Guzowski, {John F.} and Christopher Bjornsson and William Shain and Badrinath Roysam",
year = "2007",
month = "9",
doi = "10.1002/cyto.a.20430",
language = "English (US)",
volume = "71",
pages = "724--736",
journal = "Cytometry. Part A : the journal of the International Society for Analytical Cytology",
issn = "1552-4922",
publisher = "Wiley-Liss Inc.",
number = "9",

}

TY - JOUR

T1 - A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images

AU - Lin, Gang

AU - Chawla, Monica K.

AU - Olson, Kathy

AU - Barnes, Carol A

AU - Guzowski, John F.

AU - Bjornsson, Christopher

AU - Shain, William

AU - Roysam, Badrinath

PY - 2007/9

Y1 - 2007/9

N2 - Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model-based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user-provided training examples. Starting with an initial over-segmentation produced by a gradient-weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7% and classification accuracy was 93.5%, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi-model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.

AB - Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model-based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user-provided training examples. Starting with an initial over-segmentation produced by a gradient-weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7% and classification accuracy was 93.5%, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi-model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.

KW - 3D confocal microscopy

KW - Batch processing

KW - Bayesian estimator

KW - Cell nuclei

KW - Classification

KW - Model-based

KW - Parzen window

KW - Region merging

KW - Segmentation

KW - Watershed algorithm

UR - http://www.scopus.com/inward/record.url?scp=34548299771&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548299771&partnerID=8YFLogxK

U2 - 10.1002/cyto.a.20430

DO - 10.1002/cyto.a.20430

M3 - Article

C2 - 17654650

AN - SCOPUS:34548299771

VL - 71

SP - 724

EP - 736

JO - Cytometry. Part A : the journal of the International Society for Analytical Cytology

JF - Cytometry. Part A : the journal of the International Society for Analytical Cytology

SN - 1552-4922

IS - 9

ER -