Skull extraction from MR images generated by ultra short TE sequence

Mohamad Habes, Elena Rota Kops, Hans Gerd Lipinski, Hans Herzog

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

We developed two methods for the virtual extraction of the skull from the ultra short echo time MR images: i) an interactive and a semi-automatic scatterplot based segmentation as well as ii) a support vector machine (SVM) based segmentation. Both interactive and semiautomated procedures allow for good segmentation results. On the other hand it was possible to full automate the skull segmentation process with the SVM which delivered slightly better results. Four datasets were evaluated with the corresponding registered CT images using the Dice coefficients (D). The interactive scatterplot based method reached a mean D of 0.802 ± 0.070, the semi automatic one yielded a mean D of 0.791 ± 0.042 and the SVM based segmentation delivered a mean D of 0.828 ± 0.053.

Original languageEnglish (US)
Title of host publicationBildverarbeitung fur die Medizin 2012
Subtitle of host publicationAlgorithmen - Systeme - Anwendungen, BVM 2012 - Proceedings des Workshops
Pages268-273
Number of pages6
DOIs
StatePublished - Dec 1 2012
Externally publishedYes
EventWorkshops Bildverarbeitung fur die Medizin: Algorithmen - Systeme - Anwendungen, BVM 2012 - Workshop on Image Processing for Medicine: Algorithms - Systems - Applications, BVM 2012 - Berlin, Germany
Duration: Mar 18 2012Mar 20 2012

Publication series

NameInformatik aktuell
ISSN (Print)1431-472X

Conference

ConferenceWorkshops Bildverarbeitung fur die Medizin: Algorithmen - Systeme - Anwendungen, BVM 2012 - Workshop on Image Processing for Medicine: Algorithms - Systems - Applications, BVM 2012
CountryGermany
CityBerlin
Period3/18/123/20/12

ASJC Scopus subject areas

  • Modeling and Simulation

Fingerprint Dive into the research topics of 'Skull extraction from MR images generated by ultra short TE sequence'. Together they form a unique fingerprint.

Cite this