Individualizing Neuroanatomical Atlases Using a Massively Parallel Computer

  • Authors:
  • Gary E. Christensen;Michael I. Miller;Michael W. Vannier;Ulf Grenander

  • Affiliations:
  • -;-;-;-

  • Venue:
  • Computer
  • Year:
  • 1996

Quantified Score

Hi-index 4.10

Visualization

Abstract

Generalized electronic atlases of the head, created from MRI scans and labeled by experts, are currently available, but they do not allow comparison between individuals or across populations without manual input or gross simplification. The algorithms in this research overcome those problems; they create individualized, subject-specific atlases. Once the deformable atlas is individualized, it can be used to analyze data sets by querying for regions of interest, structure volumes, locations and magnitudes of abnormalities, and so on. The inverse transformation can also be used to map data from the individual's coordinate system back to the atlas' standard coordinate system. Transforming data sets to the shape of the atlas anatomy removes individual anatomical shape variations, which is useful in comparing brain function between individuals.Synthesizing individualized atlases that accurately accommodate local anatomical shape variability has many exciting potential applications.