DTI-DROID: Diffusion tensor imaging-deformable registration using orientation and intensity descriptors

Madhura Ingalhalikar, Jinzhong Yang, Christos Davatzikos, Ragini Verma

Research output: Contribution to journalArticlepeer-review

Abstract

This article presents a method (DROID) for deformable registration of diffusion tensor (DT) images that utilizes the full tensor information by integrating the intensity and orientation features into a hierarchical matching framework. The intensity features are derived from eigen value based measures that characterize the tensor in terms of its different shape properties, such as, prolateness, oblateness, and sphericity of the tensor. Local spatial distributions of the prolate, oblate, and spherical geometry are used to create an attribute vector called the geometric/intensity feature for matching. The orientation features are the orientation histograms computed from the eigenvectors. These intensity and orientation features are incorporated into a hierarchical deformable registration framework to develop a deformable registration algorithm for DT images. Using orientation features improves the matching of the white matter fiber tracts by taking into account the underlying fiber orientation information. Extensive experiments on simulated and real brain DT data show promising results that makes DROID potentially useful for subsequent group-based analysis of DT images to identify disease-induced and developmental changes in a population.

Original languageEnglish (US)
Pages (from-to)99-107
Number of pages9
JournalInternational Journal of Imaging Systems and Technology
Volume20
Issue number2
DOIs
StatePublished - Jun 1 2010

Keywords

  • Deformable registration
  • Diffusion tensor
  • Intensity descriptors
  • Orientation descriptors

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Software
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'DTI-DROID: Diffusion tensor imaging-deformable registration using orientation and intensity descriptors'. Together they form a unique fingerprint.

Cite this