Unsupervised segmentation of brain regions with similar microstructural properties: application to alcoholism

Annu Int Conf IEEE Eng Med Biol Soc. 2013:2013:1053-6. doi: 10.1109/EMBC.2013.6609685.

Abstract

In this work, a novel brain MRI segmentation approach evaluates microstructural differences between groups. Going further from the traditional segmentation of brain tissues (white matter -WM-, gray matter -GM- and cerebrospinal fluid -CSF- or a mixture of them), a new way to classify brain areas is proposed using their microstructural MR properties. Eight rats were studied using the proposed methodology identifying regions which present microstructural differences as a consequence on one month of hard alcohol consumption. Differences in relaxation times of the tissues have been found in different brain regions (p<0.05). Furthermore, these changes allowed the automatic classification of the animals based on their drinking history (hit rate of 93.75 % of the cases).

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Alcoholism / pathology*
  • Animals
  • Bayes Theorem
  • Brain / pathology*
  • Image Processing, Computer-Assisted*
  • Magnetic Resonance Imaging
  • Male
  • Rats