A hybrid approach to automatic clustering of white matter fibers

Hai Li, Zhong Xue, Lei Guo, Tianming Liu, Jill Hunter, Stephen T.C. Wong

Research output: Contribution to journalArticle

60 Scopus citations

Abstract

Recently, the tract-based white matter (WM) fiber analysis has been recognized as an effective framework to study the diffusion tensor imaging (DTI) data of human brain. This framework can provide biologically meaningful results and facilitate the tract-based comparison across subjects. However, due to the lack of quantitative definition of WM bundle boundaries, the complexity of brain architecture and the variability of WM shapes, clustering WM fibers into anatomically meaningful bundles is nontrivial. In this paper, we propose a hybrid top-down and bottom-up approach for automatic clustering and labeling of WM fibers, which utilizes both brain parcellation results and similarities between WM fibers. Our experimental results show reasonably good performance of this approach in clustering WM fibers into anatomically meaningful bundles.

Original languageEnglish (US)
Pages (from-to)1249-1258
Number of pages10
JournalNeuroImage
Volume49
Issue number2
DOIs
StatePublished - Jan 15 2010

ASJC Scopus subject areas

  • Neurology
  • Cognitive Neuroscience

Fingerprint Dive into the research topics of 'A hybrid approach to automatic clustering of white matter fibers'. Together they form a unique fingerprint.

Cite this