Split-and-merge segmentation of magnetic resonance medical images

performance evaluation and extension to three dimensions

I N Manousakas, P E Undrill, G. G. Cameron, T W Redpath

Research output: Contribution to journalArticle

62 Citations (Scopus)

Abstract

Intensity- or edge-based methods of segmentation are often insufficiently robust to be applied to images containing complex anatomical objects, such as those seen in high-resolution magnetic resonance imaging systems. Split-and-merge techniques attempt to overcome these difficulties by using homogeneity measures. Simple modifications to the basic 2D split-and-merge method, based on the principles of simulated annealing and controlled boundary elimination, are developed and discussed. Simulated annealing reduced the number of regions by 22% with a further reduction of 21% achieved through boundary elimination. Smoother regional boundaries are also produced. These methods are extended to true 3D and quantitatively compared with their 2D counterparts. The main advantage of 3D methods is that they produce segmented volumes by directly preserving the connectivity between slices, whereas in 2D, segments have to be grouped together in a post-split-and-merge process. Finally, the properties of the 3D approach are demonstrated by the automatic quantitation of brain ventricle volume, producing estimates to within 7% of validated manual methods.

Original languageEnglish
Pages (from-to)393-412
Number of pages20
JournalComputers and Biomedical Research
Volume31
Issue number6
DOIs
Publication statusPublished - Dec 1998

Keywords

  • Algorithms
  • Brain/anatomy & histology
  • Cerebral Ventricles/anatomy & histology
  • Evaluation Studies as Topic
  • Humans
  • Image Enhancement
  • Image Processing, Computer-Assisted/methods
  • Magnetic Resonance Imaging/methods
  • Reproducibility of Results
  • Software

Cite this

Split-and-merge segmentation of magnetic resonance medical images : performance evaluation and extension to three dimensions. / Manousakas, I N; Undrill, P E; Cameron, G. G.; Redpath, T W.

In: Computers and Biomedical Research, Vol. 31, No. 6, 12.1998, p. 393-412.

Research output: Contribution to journalArticle

@article{fc62ed08764749b78c29ccb7d27432e9,
title = "Split-and-merge segmentation of magnetic resonance medical images: performance evaluation and extension to three dimensions",
abstract = "Intensity- or edge-based methods of segmentation are often insufficiently robust to be applied to images containing complex anatomical objects, such as those seen in high-resolution magnetic resonance imaging systems. Split-and-merge techniques attempt to overcome these difficulties by using homogeneity measures. Simple modifications to the basic 2D split-and-merge method, based on the principles of simulated annealing and controlled boundary elimination, are developed and discussed. Simulated annealing reduced the number of regions by 22{\%} with a further reduction of 21{\%} achieved through boundary elimination. Smoother regional boundaries are also produced. These methods are extended to true 3D and quantitatively compared with their 2D counterparts. The main advantage of 3D methods is that they produce segmented volumes by directly preserving the connectivity between slices, whereas in 2D, segments have to be grouped together in a post-split-and-merge process. Finally, the properties of the 3D approach are demonstrated by the automatic quantitation of brain ventricle volume, producing estimates to within 7{\%} of validated manual methods.",
keywords = "Algorithms, Brain/anatomy & histology, Cerebral Ventricles/anatomy & histology, Evaluation Studies as Topic, Humans, Image Enhancement, Image Processing, Computer-Assisted/methods, Magnetic Resonance Imaging/methods, Reproducibility of Results, Software",
author = "Manousakas, {I N} and Undrill, {P E} and Cameron, {G. G.} and Redpath, {T W}",
note = "Copyright 1998 Academic Press.",
year = "1998",
month = "12",
doi = "10.1006/cbmr.1998.1489",
language = "English",
volume = "31",
pages = "393--412",
journal = "Computers and Biomedical Research",
issn = "0010-4809",
publisher = "Academic Press Inc.",
number = "6",

}

TY - JOUR

T1 - Split-and-merge segmentation of magnetic resonance medical images

T2 - performance evaluation and extension to three dimensions

AU - Manousakas, I N

AU - Undrill, P E

AU - Cameron, G. G.

AU - Redpath, T W

N1 - Copyright 1998 Academic Press.

PY - 1998/12

Y1 - 1998/12

N2 - Intensity- or edge-based methods of segmentation are often insufficiently robust to be applied to images containing complex anatomical objects, such as those seen in high-resolution magnetic resonance imaging systems. Split-and-merge techniques attempt to overcome these difficulties by using homogeneity measures. Simple modifications to the basic 2D split-and-merge method, based on the principles of simulated annealing and controlled boundary elimination, are developed and discussed. Simulated annealing reduced the number of regions by 22% with a further reduction of 21% achieved through boundary elimination. Smoother regional boundaries are also produced. These methods are extended to true 3D and quantitatively compared with their 2D counterparts. The main advantage of 3D methods is that they produce segmented volumes by directly preserving the connectivity between slices, whereas in 2D, segments have to be grouped together in a post-split-and-merge process. Finally, the properties of the 3D approach are demonstrated by the automatic quantitation of brain ventricle volume, producing estimates to within 7% of validated manual methods.

AB - Intensity- or edge-based methods of segmentation are often insufficiently robust to be applied to images containing complex anatomical objects, such as those seen in high-resolution magnetic resonance imaging systems. Split-and-merge techniques attempt to overcome these difficulties by using homogeneity measures. Simple modifications to the basic 2D split-and-merge method, based on the principles of simulated annealing and controlled boundary elimination, are developed and discussed. Simulated annealing reduced the number of regions by 22% with a further reduction of 21% achieved through boundary elimination. Smoother regional boundaries are also produced. These methods are extended to true 3D and quantitatively compared with their 2D counterparts. The main advantage of 3D methods is that they produce segmented volumes by directly preserving the connectivity between slices, whereas in 2D, segments have to be grouped together in a post-split-and-merge process. Finally, the properties of the 3D approach are demonstrated by the automatic quantitation of brain ventricle volume, producing estimates to within 7% of validated manual methods.

KW - Algorithms

KW - Brain/anatomy & histology

KW - Cerebral Ventricles/anatomy & histology

KW - Evaluation Studies as Topic

KW - Humans

KW - Image Enhancement

KW - Image Processing, Computer-Assisted/methods

KW - Magnetic Resonance Imaging/methods

KW - Reproducibility of Results

KW - Software

U2 - 10.1006/cbmr.1998.1489

DO - 10.1006/cbmr.1998.1489

M3 - Article

VL - 31

SP - 393

EP - 412

JO - Computers and Biomedical Research

JF - Computers and Biomedical Research

SN - 0010-4809

IS - 6

ER -