Format

Send to:

Choose Destination
See comment in PubMed Commons below
PLoS One. 2012;7(9):e43415. doi: 10.1371/journal.pone.0043415. Epub 2012 Sep 6.

Improving DTI tractography by including diagonal tract propagation.

Author information

  • 1Department of Radiology, UMDNJ-New Jersey Medical School, Newark, New Jersey, United States of America. neon.taylor@gmail.com

Abstract

Tractography algorithms have been developed to reconstruct likely WM pathways in the brain from diffusion tensor imaging (DTI) data. In this study, an elegant and simple means for improving existing tractography algorithms is proposed by allowing tracts to propagate through diagonal trajectories between voxels, instead of only rectilinearly to their facewise neighbors. A series of tests (using both real and simulated data sets) are utilized to show several benefits of this new approach. First, the inclusion of diagonal tract propagation decreases the dependence of an algorithm on the arbitrary orientation of coordinate axes and therefore reduces numerical errors associated with that bias (which are also demonstrated here). Moreover, both quantitatively and qualitatively, including diagonals decreases overall noise sensitivity of results and leads to significantly greater efficiency in scanning protocols; that is, the obtained tracts converge much more quickly (i.e., in a smaller amount of scanning time) to those of data sets with high SNR and spatial resolution. Importantly, the inclusion of diagonal propagation adds essentially no appreciable time of calculation or computational costs to standard methods. This study focuses on the widely-used streamline tracking method, FACT (fiber assessment by continuous tracking), and the modified method is termed "FACTID" (FACT including diagonals).

PMID:
22970125
[PubMed - indexed for MEDLINE]
PMCID:
PMC3435381
Free PMC Article
PubMed Commons home

PubMed Commons

0 comments
How to join PubMed Commons

    Supplemental Content

    Full text links

    Icon for Public Library of Science Icon for PubMed Central
    Loading ...
    Write to the Help Desk