Article Text

Download PDFPDF
  1. Natasha Clarke,
  2. Thomas Barrick,
  3. Peter Garrard
  1. St George's University of London


Diffusion Tensor Imaging (DTI) Segmentation (D-SEG) relies on variation in tissue microstructure characteristics of the diffusion tensor to segment regions of the brain with similar diffusion properties. Previously shown to be a reliable tool for parcelating and classifying volumes of interest (VOIs) within brain tumours, generating objective tissue type boundaries (Jones et al., 2014), it also relates to differences in age, executive function and working memory (Williams et al., unpublished data). We applied D-SEG to a novel data set comprising 18 DTI scans; 9 controls and 9 patients with clinically and/or radiologically confirmed focal dementia syndromes: 4 Progressive non-fluent Aphasia (PnFA), 2 Semantic Dementia (SD) and 3 Logopenic Aphasia (LPA). The D-SEG clustering technique takes isotropic (p) and anisotropic (q) diffusion values plotted in a 2D space and uses iterative k-means clustering analysis to segment the images into areas representing tissue with similar diffusion properties. We found a statistical difference between patients and control groups in 2 of the segments, specifically grey matter and well organised white matter. Greater grey matter was positively correlated with scores on the Mini Linguistic State Examination (MLSE). These findings add to the growing body of work supporting D-SEG as a valid tool for image analysis in clinical populations.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.