Robust segmentation of brain MRI using combination of registration and EM-based methods

Maria Murgasova, Mary Rutherford, Leigh Dyet, David Edwards, Jo Hajnal, Daniel Rueckert

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

Brain MRI segmentation is usually performed either based on intensity or non-rigid alignment with manually segmented image. Both of these approaches have drawbacks. Registration-based methods cannot often capture the complexity of cortical folding. On the other hand intensity-based methods assume homogeneous intensities of the tissues which is not well aligned with natural intensity variations of brain tissues. We combine both approaches to obtain a robust segmentation method which is successfully used on healthy and diseased brain MRI of 2-year-old children..

Original languageEnglish
Title of host publicationProceedings - SCCG 2007
Subtitle of host publication23rd Spring Conference on Computer Graphics
EditorsStephen N. Spencer
PublisherAssociation for Computing Machinery, Inc
Pages107-114
Number of pages8
ISBN (Electronic)9781605589565
DOIs
StatePublished - 2007
Externally publishedYes
Event23rd Spring Conference on Computer Graphics, SCCG 2007 - Budmerice, Slovakia
Duration: 26 Apr 200728 Apr 2007

Publication series

NameProceedings - SCCG 2007: 23rd Spring Conference on Computer Graphics

Conference

Conference23rd Spring Conference on Computer Graphics, SCCG 2007
Country/TerritorySlovakia
CityBudmerice
Period26/04/0728/04/07

Keywords

  • Brain MRI
  • Expectation-maximization
  • Non-rigid registration
  • Segmentation

Fingerprint

Dive into the research topics of 'Robust segmentation of brain MRI using combination of registration and EM-based methods'. Together they form a unique fingerprint.

Cite this