The mirror method of assessing segmentation quality in atlas label propagation

R. A. Heckemann, A. Hammers, P. Aljabar, D. Rueckert, J. V. Hajnal

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

4 Scopus citations

Abstract

Atlas-based brain image segmentation quality is difficult to assess in the absence of reference target segmentations. We propose a measure of segmentation success based on transforming the atlas label twice: once by registering the atlas to the target and a second time by registering the target to the atlas. Each registration represents transformations by free-form deformations. The overlap between the twice-transformed label and the original ('mirror overlap') correlates with the forward overlap (between the once-transformed label and a target reference), especially for subcortical structures. Using mirror overlap as an atlas selection criterion results in improved segmentations versus random selection.

Original languageEnglish
Title of host publicationProceedings - 2009 IEEE International Symposium on Biomedical Imaging
Subtitle of host publicationFrom Nano to Macro, ISBI 2009
Pages1194-1197
Number of pages4
DOIs
StatePublished - 2009
Externally publishedYes
Event2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2009 - Boston, MA, United States
Duration: 28 Jun 20091 Jul 2009

Publication series

NameProceedings - 2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2009

Conference

Conference2009 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2009
Country/TerritoryUnited States
CityBoston, MA
Period28/06/091/07/09

Keywords

  • Atlas
  • Brain
  • Image segmentation
  • Magnetic resonance imaging

Fingerprint

Dive into the research topics of 'The mirror method of assessing segmentation quality in atlas label propagation'. Together they form a unique fingerprint.

Cite this