Local and global volume changes of subcortical brain structures from longitudinally varying neuroimaging data for dementia identification

Comput Med Imaging Graph. 2012 Sep;36(6):464-73. doi: 10.1016/j.compmedimag.2012.03.006. Epub 2012 May 31.


Quantification of structural changes in the human brain is important to elicit resemblances and differences between pathological and normal aging. Identification of dementia, associated with loss of cognitive ability beyond normal aging, and especially converters--the subgroup of individuals at risk for developing dementia--has recently gained importance. For this purpose atrophy markers have been explored and their effectiveness has been evaluated both cross-sectionally and longitudinally. However, more research is needed to understand the dynamics of atrophy markers at different disease stages, which requires temporal analysis of local along with global changes. Unfortunately, most of the longitudinal neuroimaging data available in the clinical settings is acquired at largely varying time intervals. In the light of the above, this study presents a novel methodology to process longitudinal neuroimaging data acquired incompletely and at different time intervals, and explores complementary nature of local and global brain volume changes in identifying dementia. Results on the OASIS database demonstrate discriminative power of global atrophy in hippocampus (as early as two years after the first visit) for identifying demented cases, and local volume shrinkage of thalamus proper (as early as three years after the first visit) for differentiating converters.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Algorithms
  • Brain / pathology*
  • Dementia / pathology*
  • Humans
  • Image Enhancement / methods
  • Image Interpretation, Computer-Assisted / methods*
  • Imaging, Three-Dimensional / methods*
  • Longitudinal Studies
  • Magnetic Resonance Imaging / methods*
  • Middle Aged
  • Neuroimaging / methods*
  • Organ Size
  • Reproducibility of Results
  • Sensitivity and Specificity
  • Subtraction Technique*