Note on Indices of Shape and Similarity for Diffusion Tensors
D. Alexander. Research Notes, Department of Computer Science, UCL (University College London), (October 2000)
Abstract
A novel approach to the problem of finding appropriate similarity
measures for
DT-MRI data is presented. There are several existing measures in the
literature,
but there is little indication of which measures are appropriate for
particular
applications. A mathematical framework is introduced here, based on
the physics
of the diffusion tensor, which regards the tensor as a real valued
function of the
unit sphere in 3D. Given two functions of the same domain, their total
difference
(or similarity) can be expressed as the integral of their difference
over their
domain. This approach is applied to diffusion tensors and a family
of cumulative
tensor difference measures is derived. Effectiveness of the resulting
measures is
validated within an image registration application. Furthermore, this
framework
can be used to derive and alternative set of DT size and shape indices,
which are
analogous to existing measures and are presented here.
%0 Report
%1 Alexander2000
%A Alexander, Daniel
%D 2000
%K Diffusion, Tensor DTI Imaging, Diffusion
%T Note on Indices of Shape and Similarity for Diffusion Tensors
%X A novel approach to the problem of finding appropriate similarity
measures for
DT-MRI data is presented. There are several existing measures in the
literature,
but there is little indication of which measures are appropriate for
particular
applications. A mathematical framework is introduced here, based on
the physics
of the diffusion tensor, which regards the tensor as a real valued
function of the
unit sphere in 3D. Given two functions of the same domain, their total
difference
(or similarity) can be expressed as the integral of their difference
over their
domain. This approach is applied to diffusion tensors and a family
of cumulative
tensor difference measures is derived. Effectiveness of the resulting
measures is
validated within an image registration application. Furthermore, this
framework
can be used to derive and alternative set of DT size and shape indices,
which are
analogous to existing measures and are presented here.
@techreport{Alexander2000,
abstract = {A novel approach to the problem of finding appropriate similarity
measures for
DT-MRI data is presented. There are several existing measures in the
literature,
but there is little indication of which measures are appropriate for
particular
applications. A mathematical framework is introduced here, based on
the physics
of the diffusion tensor, which regards the tensor as a real valued
function of the
unit sphere in 3D. Given two functions of the same domain, their total
difference
(or similarity) can be expressed as the integral of their difference
over their
domain. This approach is applied to diffusion tensors and a family
of cumulative
tensor difference measures is derived. Effectiveness of the resulting
measures is
validated within an image registration application. Furthermore, this
framework
can be used to derive and alternative set of DT size and shape indices,
which are
analogous to existing measures and are presented here.},
added-at = {2007-01-10T11:43:56.000+0100},
author = {Alexander, Daniel},
biburl = {https://www.bibsonomy.org/bibtex/2643bee0a7cb87bdc498df7c6082f3168/bmeyer},
description = {Diffusion Tensor Imaging (DTI)},
institution = {Department of Computer Science, UCL (University College London)},
interhash = {1732e92b95a00e4ee7e8c32fd0710c67},
intrahash = {643bee0a7cb87bdc498df7c6082f3168},
keywords = {Diffusion, Tensor DTI Imaging, Diffusion},
month = {October},
pdf = {http://www.cs.ucl.ac.uk/brandnew/research2/Publications/rn14.pdf},
timestamp = {2007-01-10T11:43:56.000+0100},
title = {Note on Indices of Shape and Similarity for Diffusion Tensors},
type = {Research Notes},
year = 2000
}