Publication
Title
Automated correction of improperly rotated diffusion gradient orientations in diffusion weighted MRI
Author
Abstract
Ensuring one is using the correct gradient orientations in a diffusion MRI study can be a challenging task. As different scanners, file formats and processing tools use different coordinate frame conventions, in practice, users can end up with improperly oriented gradient orientations. Using such wrongly oriented gradient orientations for subsequent diffusion parameter estimation will invalidate all rotationally variant parameters and fiber tractography results. While large misalignments can be detected by visual inspection, small rotations of the gradient table (e.g. due to angulation of the acquisition plane), are much more difficult to detect. In this work, we propose an automated method to align the coordinate frame of the gradient orientations with that of the corresponding diffusion weighted images, using a metric based on whole brain fiber tractography. By transforming the gradient table and measuring the average fiber trajectory length, we search for the transformation that results in the best global 'connectivity'. To ensure a fast calculation of the metric we included a range of algorithmic optimizations in our tractography routine. To make the optimization routine robust to spurious local maxima, we use a stochastic optimization routine that selects a random set of seed points on each evaluation. Using simulations, we show that our method can recover the correct gradient orientations with high accuracy and precision. In addition, we demonstrate that our technique can successfully recover rotated gradient tables on a wide range of clinically realistic data sets. As such, our method provides a practical and robust solution to an often overlooked pitfall in the processing of diffusion MRI. (C) 2014 Elsevier B.V. All rights reserved.
Language
English
Source (journal)
Medical image analysis. - Amsterdam
Publication
Amsterdam : 2014
ISSN
1361-8415
Volume/pages
18:7(2014), p. 953-962
ISI
000342256200002
Full text (Publisher's DOI)
Full text (publisher's version - intranet only)
UAntwerpen
Faculty/Department
Research group
[E?say:metaLocaldata.cgzprojectinf]
Integrated cerebral networks for perception, cognition and action in human and non-human primates (CEREBNET).
Publication type
Subject
Affiliation
Publications with a UAntwerp address
External links
Web of Science
Record
Identification
Creation 06.11.2014
Last edited 04.11.2017
To cite this reference