Localized components analysis

  • Authors:

  • Dan Alcantara;Owen Carmichael;Eric Delson;Will Harcourt-Smith;Kirsten Sterner;Stephen Frost;Rebecca Dutton;Paul Thompson;Howard Aizenstein;Oscar Lopez;James Becker;Nina Amenta

  • Affiliations:

  • Computer Science Departments, University of California, Davis;Computer Science Departments, University of California, Davis and Neurology Departments, University of California, Davis;Lehman College of the City University of New York and NYCEP, American Museum of Natural History;Lehman College of the City University of New York and NYCEP, American Museum of Natural History;NYCEP, American Museum of Natural History;Anthropology Department, University of Oregon;Neurology Department and Laboratory of Neuro Imaging, University of California, Los Angeles;Neurology Department and Laboratory of Neuro Imaging, University of California, Los Angeles;Psychiatry Departments, University of Pittsburgh;Neurology, University of Pittsburgh;Psychiatry Departments, University of Pittsburgh and Neurology Departments, University of Pittsburgh and Psychology Departments, University of Pittsburgh;Computer Science Departments, University of California, Davis

  • Venue:
  • IPMI'07 Proceedings of the 20th international conference on Information processing in medical imaging
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

We introduce Localized Components Analysis (LoCA) for describing surface shape variation in an ensemble of biomedical objects using a linear subspace of spatially localized shape components. In contrast to earlier methods, LoCA optimizes explicitly for localized components and allows a flexible trade-off between localized and concise representations. Experiments comparing LoCA to a variety of competing shape representation methods on 2D and 3D shape ensembles establish the superior ability of LoCA to modulate the locality-conciseness trade-off and generate shape components corresponding to intuitive modes of shape variation. Our formulation of locality in terms of compatibility between pairs of surface points is shown to be flexible enough to enable spatially-localized shape descriptions with attractive higher-order properties such as spatial symmetry.