Meeting Banner
Abstract #0777

Atlas-Guided Cluster Analysis of Fiber Tracts

Christian Ros1, 2, Daniel Guellmar1, Martin Stenzel2, Hans-Joachim Mentzel2, Jrgen R. Reichenbach3

1Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena , Jena, TH, Germany; 2Pediatric Radiology, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena , Jena, TH, Germany; 3Medical Physics Group, Institute of Diagnostic and Interventional Radiology I, Jena University Hospital - Friedrich Schiller University Jena, Jena, TH, Germany

With this contribution we present a new hybrid approach that incorporates anatomic information of a probabilistic white matter fiber bundle atlas into the cluster analysis. This technique enables the robust and consistent extraction of fiber bundles that correspond to the classes in the atlas. In addition, it offers the advantage to identify bundles in the data set that are not defined in the atlas. To validate the robustness and the consistent extraction for multiple subjects, data sets of 46 healthy subjects were processed and atlas-guided clustering was performed.

Abstract PDF Presentation Video

Keywords

addition advance analyses analyzing anatomical anisotropy apparent array assessed atlas atlases audience automatic automatically belong blue body brain bundle bundles class classes clinical cluster clustered clustering clusters coded color colors consisted consistent constructed consuming contains correct correspond criterion datasets delineation determine diagnostic diffusion diffusivity direct distance distances education efficient eminently empty enable enables even evolved example exemplary extended extract extracted extraction facilitates federal fiber five forceps fractional framework fulfilled fully generation gradient guided hence highest highly hospital hybrid identification identify important incorporated incorporates indicated indicates individual influence institute integrate integrating intensity kind length linear major materials matrix medical membership memberships merging minor modulates part pediatric performance planar population predefined previously probabilistic probabilities probability process project quantitative radiology reconstructed reduced reliability reliable representation repulsion scanner selection sets shortest slice space sqrt stopping structures subject subjects subsequently successfully successively suitable superior support target temporal tracts twice upon various visualized volunteers want white whole