PLoS ONE (Jan 2013)

Diffusion tensor tractography versus volumetric imaging in the diagnosis of behavioral variant frontotemporal dementia.

  • Alexander Frizell Santillo,
  • Johanna Mårtensson,
  • Olof Lindberg,
  • Markus Nilsson,
  • Amir Manzouri,
  • Maria Landqvist Waldö,
  • Danielle van Westen,
  • Lars-Olof Wahlund,
  • Jimmy Lätt,
  • Christer Nilsson

DOI
https://doi.org/10.1371/journal.pone.0066932
Journal volume & issue
Vol. 8, no. 7
p. e66932

Abstract

Read online

MRI diffusion tensor imaging (DTI) studies of white matter integrity in behavioral variant frontotemporal dementia have consistently shown involvement of frontal and temporal white matter, corresponding to regional loss of cortical volume. Volumetric imaging has a suboptimal sensitivity as a diagnostic tool and thus we wanted to explore if DTI is a better method to discriminate patients and controls than volumetric imaging. We examined the anterior cingulum bundle in 14 patients with behavioral variant frontotemporal dementia and 22 healthy controls using deterministic manual diffusion tensor tractography, and compared DTI parameters with two measures of cortical atrophy, VBM and cortical thickness, of the anterior cingulate cortex (ACC). Statistically significant changes between patients and controls were detected in all DTI parameters, with large effect sizes. ROC-AUC was for the best DTI parameters: 0.92 (fractional anisotropy) to 0.97 (radial diffusivity), 0.82 for the best cortical parameter, VBM of the ACC. Results from the AUC were confirmed with binary logistic regression analysis including demographic variables, but only for fractional anisotropy and mean diffusivity. Ability to classify patient/nonpatient status was significantly better for mean diffusivity vs. VBM (p=0.031), and borderline significant for fractional anisotropy vs. VBM (p=0.062). The results indicate that DTI could offer advantages in comparison with the assessment of cortical volume in differentiating patients with behavioral variant frontotemporal dementia and controls.