Integration of fuzzy spatial relations in deformable models-Application to brain MRI segmentation

  • Authors:

  • Olivier Colliot;Oscar Camara;Isabelle Bloch

  • Affiliations:

  • ícole Nationale Supérieure des Télécommunications, Département TSI, CNRS UMR 5141 LTCI, 46, rue Barrault, 75634 Paris Cedex 13, France;ícole Nationale Supérieure des Télécommunications, Département TSI, CNRS UMR 5141 LTCI, 46, rue Barrault, 75634 Paris Cedex 13, France;ícole Nationale Supérieure des Télécommunications, Département TSI, CNRS UMR 5141 LTCI, 46, rue Barrault, 75634 Paris Cedex 13, France

  • Venue:
  • Pattern Recognition
  • Year:
  • 2006

Quantified Score

Hi-index 0.01

Visualization

Abstract

This paper presents a general framework to integrate a new type of constraints, based on spatial relations, in deformable models. In the proposed approach, spatial relations are represented as fuzzy subsets of the image space and incorporated in the deformable model as a new external force. Three methods to construct an external force from a fuzzy set representing a spatial relation are introduced and discussed. This framework is then used to segment brain subcortical structures in magnetic resonance images (MRI). A training step is proposed to estimate the main parameters defining the relations. The results demonstrate that the introduction of spatial relations in a deformable model can substantially improve the segmentation of structures with low contrast and ill-defined boundaries.