Fiber Tracking & Connectivity Mapping

Room 718 A


Chairs: Derek K. Jones and Mariana Lazar


Prog #

10:30 425. Young Investigator Award Finalist: Gibbs Tracking: A Novel Approach for the Reconstruction of Neuronal Pathways

Björn Wolf Kreher1, Irina Mader2, 3, Valerij G. Kiselev1

1University Hospital Freiburg, Freiburg, Germany; 2Neurocenter of the University Hospital Freiburg, Freiburg, Germany; 3Department of Neurology of the University Hospital Freiburg, Freiburg, Germany

The known fibre tracking methods are commonly based on the ‘walker-principle’. There, fibres are reconstructed as trajectories of ‘walkers’ which are guided either deterministically or randomly according to the local properties of the diffusion-weighted signal. In this study, a principally new method of fibre tracking is proposed which is not based on any ‘walking’ algorithm. It resolves a number of inherent problems of the ‘walking’ approach being capable to reconstruct crossing and spreading fibre configurations. The performance of this new method is shown in simulation and in an in vivo measurement

10:50 426. Probabilistic Diffusion-Tensor Fiber Tractography in a Bayesian Framework with an Atlas Prior

Philip A. Cook1, Hui Zhang1, Suyash P. Awate1, James C. Gee1

1University of Pennsylvania, Philadelphia, Pennsylvania, USA

This work uses a diffusion-tensor atlas to guide probabilistic tractography in a Bayesian framework. The spatial normalization explicitly optimizes tensor similarity to provide optimal alignment of white matter. The mean and variance of the fiber orientation in each voxel is used as a prior for the tractography method. We demonstrate the method using an atlas constructed from eleven subjects and atlas-guided tracking in one subject. Results are shown from tracking of the cingulum, formix, and mid-cerebellar peduncle. In each case, the atlas prior increases the connectivity along the pathway relative to an equivalent method without a prior.

11:02 427. Quantitative Validation of MR Tractography Using the CoCoMac Database

Patric Hagmann1, Xavier Gigandet2, Reto Meuli1, Rolf Kötter3, Olaf Sporns4, Van J. Wedeen5

1University Hospital Center and University of Lausanne, Lausanne, Switzerland; 2Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland; 3Radboud University Nijmegen Medical Centre, Nijmegen, Netherlands; 4Indiana University, Bloomington, USA; 5Harvard Medical School, Charlestowm, Massachusetts, USA

We quantitavely characterize MR tractography performance in terms of sensitivity and specificity by comparing the connection matrix of an MR tractography experiment in a fixed macaque hemisphere with a set of histologic tracing studies made in the same species that establish 313 connections as well as 469 absent connections. Computing the Receiver Operator Characteristic curve, we see that MR tractography agrees significantly with the gold standard. Typically a sensitivity of 60% for a specificity of 70% can be achieved.

11:14 428. Combining Tractography and Coherence Measures to Identify Connectivity Within a Neural Network for Reading

Stephen Welbourne1, Karl Embleton1, Hamied Haroon1, David Morris1, Geoffrey J M Parker1, Matthew Anthony Lambon Ralph1

1University of Manchester, Manchester, UK

PICo probabilistic tractography was used to investigate the anatomical connectivity of functional networks which had been suggested by two previous studies on reading. Functional connections identified by the MEG study were found to be significant predictors of the tractography results; however functional connections from the DCM study did not significantly predict the tractography results. More detailed analysis of the pattern of results suggests that while there was strong evidence supporting most of the MEG connections two of the hypothesised connections were not supported by tractography. It is suggested that tractography has a useful part to play in constraining model selection.

11:26 429. Quantification of White Matter Fiber Orientation at Tumor Margins with Diffusion Tensor Invariant Gradients

Gordon Kindlmann1, Stephen Whalen1, Ralph O. Suarez1, Alex J. Golby1, Carl-Fredrik Westin1

1Harvard Medical School, Boston, Massachusetts, USA

Existing applications of Diffusion Tensor Imaging (DTI) to neurosurgical planning of tumor resection focus on the tensor principal eigenvector (either in visualization with RGB colormaps, or connectivity analysis by tractography), or on changes to scalar-valued tensor invariants such as trace (as with edema) or fractional anisotropy (axonal destruction).  We present new metrics, Diffusion Angle (Da) and Diffusion Fraction (Df), for characterizing fiber orientation relative to affected regions, based on measuring the spatial gradients of tensor invariants.  The metrics are motivated in terms of their expected behavior along cortical gyrii, and initially evaluated on a clinical brain tumor case.

11:38 430.
 [Not Available]
Computing Strings of Average HARDI Models Using Procrustes-Based Fibre Alignment

Irina Kezele1, Cyril Poupon1, Muriel Perrin2, Yann Cointepas1, Vincent El Kouby1, Fabrice Poupon1, Jean-François Mangin1

1CEA NeuroSpin, Gif sur Yvette, France; 2GE Healthcare, Buc, France

The idea underpinning the work we present is to design robust and objective tools for brain white matter (WM) morphometry. We focus on WM tracts, and propose to represent them by their mean lines, to which we associate the attributes derived from high-angular resolution diffusion imaging (HARDI).The definition of the tract mean line derives directly from the geometry of the tract fibres. We determine the fibre point correspondences and impact factors of individual fibres, upon which we estimate average HARDI models along tract mean lines, thus obtaining a compact tract representation free of the outlier influence and undesired edge effects.

11:50 431. Software Tool to Generate Complex Structures for Validation of Fibre Tracking

Tom G. Close1, 2, J-Donald Tournier1, 3, Fernando Calamante1, 3, Leigh A. Johnston, 24, Iven Mareels, 23, Alan Connelly1, 3

1Brain Research Institute, Melbourne, Australia; 2National ICT Australia, Melbourne, Australia; 3University of Melbourne, Melbourne, Australia; 4Howard Florey Institute, Melbourne, Australia

Despite important advances in fibre-tracking algorithms, the field is currently limited by the lack of a ‘gold standard’. Current validation methods (e.g. comparisons with known anatomy, ex vivo measurements and simulated data) suffer from limitations and/or uncertainties. We have developed a fast flexible software-tool that can generate random models of fibre bundles with a complexity comparable to white matter structures. This provides an essential tool for current and future tracking algorithms to be robustly tested, and their limitations more thoroughly understood.

12:02  432. Visualisation of CSI Metabolite Concentrations Along Specific White Matter Tracts

Hannah Joan Coward1, Ruth Louise O'Gorman1, 2, David J. Lythgoe1, Marco Catani1

1Institute of Psychiatry, London, UK; 2King's College Hospital, UK

Diffusion Tensor Imaging (DTI) is a non-invasive method to visualise gross white matter architecture and probe microstructural integrity of both normal and pathological brain.  Equivalently, quantitative Magnetic Resonance Spectroscopy (MRS) is a non-invasive technique used to measure metabolic concentrations, changes of which have been documented in several neurological disorders. Here preliminary work that combines MRS Chemical Shift Imaging (CSI) with DTI-tractography is presented.  Our aim is to assess the feasibility of performing tract-specific measurements of MRS brain metabolites along the left arcuate fasciculus (AF).

12:14 433. Using Boy’s Real Projective Plane Immersion for Coloring DT-MRI Slices

Cagatay Demiralp1, John F. Hughes1, David H. Laidlaw1

1Brown University, Providence, Rhode Island, USA

We introduce Boy’s surface, an immersion of the projective plane (RP2) in 3D, as a model for mapping tensor orientation to colors. One of the most popular methods of visualizing DT-MRI slices is to map principal eigenvectors of the underlying tensors to RGB colors. All the existing variations of this method, however, suffer from mirror symmetry one way or another. Ideally, we would like to map 1) different orientations to different, unique colors and 2) similar orientations to similar colors. As trivial as it sounds, this task is not easy and there is a compelling reason for that:  The problem of smooth, one-to-one assignment of colors to orientations is tantamount to embedding the real projective plane in 3D, which was proven to be impossible.  While the real projective plane does not have an embedding in 3D, it has immersions. Here, we present Boy’s immersion of the real projective plane for mapping orientation to colors. Boy’s surface is the only real projective plane immersion without singularities. Coloring based on this model (we refer it as RP2 coloring) will map different orientations to different and similar colors, except along the self-intersection curve.