3D tensor normalization for improved accuracy in DTI tensor registration methods

Aditya Gupta, Maria Escolar, Cheryl Dietrich, John Gilmore, Guido Gerig, Martin Styner

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper presents a method for normalization of diffusion tensor images (DTI) to a fixed DTI template, a pre-processing step to improve the performance of full tensor based registration methods. The proposed method maps the individual tensors of the subject image in to the template space based on matching the cumulative distribution function and the fractional anisotrophy values. The method aims to determine a more accurate deformation field from any full tensor registration method by applying the registration algorithm on the normalized DTI rather than the original DTI. The deformation field applied to the original tensor images are compared to the deformed image without normalization for 11 different cases of mapping seven subjects (neonate through 2 years) to two different atlases. The method shows an improvement in DTI registration based on comparing the normalized fractional anisotropy values of major fiber tracts in the brain.

Original languageEnglish (US)
Title of host publicationBiomedical Image Registration - 5th International Workshop, WBIR 2012, Proceedings
Pages170-179
Number of pages10
DOIs
StatePublished - 2012
Event5th International Workshop on Biomedical Image Registration, WBIR 2012 - Nashville, TN, United States
Duration: Jul 7 2012Jul 8 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7359 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other5th International Workshop on Biomedical Image Registration, WBIR 2012
Country/TerritoryUnited States
CityNashville, TN
Period7/7/127/8/12

Keywords

  • DTI Registration
  • DTITK
  • Tensor Normalization

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of '3D tensor normalization for improved accuracy in DTI tensor registration methods'. Together they form a unique fingerprint.

Cite this