Difference between revisions of "Projects:QuantitativeSusceptibilityMapping"

From NAMIC Wiki
Jump to: navigation, search
Line 16: Line 16:
 
= Description =
 
= Description =
  
There is increasing evidence that excessive iron deposition in specifi�c regions
+
There is increasing evidence that excessive iron deposition in specific regions
 
of the brain is associated with neurodegenerative disorders such as Alzheimer's
 
of the brain is associated with neurodegenerative disorders such as Alzheimer's
 
and Parkinson's disease [1]. The role of iron in the pathogenesis of these diseases
 
and Parkinson's disease [1]. The role of iron in the pathogenesis of these diseases
Line 22: Line 22:
 
to quantify its concentration in-vivo. Since iron is a ferromagnetic substance,
 
to quantify its concentration in-vivo. Since iron is a ferromagnetic substance,
 
changes in iron concentration result in local changes in the magnetic suscepti-
 
changes in iron concentration result in local changes in the magnetic suscepti-
bility of tissue. In magnetic resonance imaging (MRI) experiments, di�fferences
+
bility of tissue. In magnetic resonance imaging (MRI) experiments, differences
in magnetic susceptibility cause perturbations in the local magnetic fi�eld, which
+
in magnetic susceptibility cause perturbations in the local magnetic field, which
 
can be computed from the phase of the MR signal.
 
can be computed from the phase of the MR signal.
The advent of high �field MRI has made it possible for subtle diff�erences in
+
The advent of high field MRI has made it possible for subtle differences in
susceptibility to measurably aff�ect the phase of the MR signal. This has raised
+
susceptibility to measurably affect the phase of the MR signal. This has raised
the possibility that estimating magnetic susceptibility from high �field phase data
+
the possibility that estimating magnetic susceptibility from high field phase data
 
may provide a new means for quantifying iron concentration in-vivo.
 
may provide a new means for quantifying iron concentration in-vivo.
  
The �field perturbations caused by magnetic susceptibility diff�erences can be
+
The field perturbations caused by magnetic susceptibility differences can be
 
modeled as the convolution of a dipole-like kernel with the spatial susceptibility
 
modeled as the convolution of a dipole-like kernel with the spatial susceptibility
 
distribution. In the Fourier domain, the kernel exhibits zeros at the magic angle,
 
distribution. In the Fourier domain, the kernel exhibits zeros at the magic angle,
 
preventing direct inversion of the fieldmap [3]. Critically, limited observations of
 
preventing direct inversion of the fieldmap [3]. Critically, limited observations of
the fi�eld make the problem ill-posed. The observed data is also corrupted by con-
+
the field make the problem ill-posed. The observed data is also corrupted by con-
 
founding biasfields (ie. those from tissue-air interfaces, mis-set shims, and other
 
founding biasfields (ie. those from tissue-air interfaces, mis-set shims, and other
non-local sources). Eliminating these fi�elds is critical for accurate susceptibility
+
non-local sources). Eliminating these fields is critical for accurate susceptibility
 
estimation since they corrupt the phase contributions from local susceptibility
 
estimation since they corrupt the phase contributions from local susceptibility
 
sources.
 
sources.
  
 
In general, methods that rely heavily on agreement between observed and
 
In general, methods that rely heavily on agreement between observed and
predicted �field values computed using kernel-based forward models [2, 3, 6] are in-
+
predicted field values computed using kernel-based forward models [2, 3, 6] are in-
herently limited since they cannot distinguish between low frequency biasfi�elds
+
herently limited since they cannot distinguish between low frequency biasfields
 
and susceptibility distributions that are eigenfunctions of the model. Examples
 
and susceptibility distributions that are eigenfunctions of the model. Examples
 
of such distributions include constant, linear, and quadratic functions of suscep-
 
of such distributions include constant, linear, and quadratic functions of suscep-
tibility along the main �field (ie. 'z') direction. Applying the forward model to
+
tibility along the main field (ie. 'z') direction. Applying the forward model to
these distributions results in predicted �fields that are proportional to the local
+
these distributions results in predicted fields that are proportional to the local
susceptibility sources, but also identical in form to non-local bias�fields (ie. those
+
susceptibility sources, but also identical in form to non-local biasfields (ie. those
produced by a z-shim). Therefore, removing all low frequency �fields prior to
+
produced by a z-shim). Therefore, removing all low frequency fields prior to
susceptibility estimation will eliminate the bias�field as well as �fields due to the
+
susceptibility estimation will eliminate the biasfield as well as fields due to the
 
sources of interest, potentially preventing accurate calculation of the underlying
 
sources of interest, potentially preventing accurate calculation of the underlying
susceptibility values. In contrast, inadequate removal of the bias�field may result
+
susceptibility values. In contrast, inadequate removal of the biasfield may result
 
in the estimation of artifactual susceptibility eigenfunctions in areas where the
 
in the estimation of artifactual susceptibility eigenfunctions in areas where the
bias�field is strong, such as regions adjacent to tissue-air interfaces. This suggests
+
biasfield is strong, such as regions adjacent to tissue-air interfaces. This suggests
 
that additional information such as boundary conditions or priors may be neces-
 
that additional information such as boundary conditions or priors may be neces-
 
sary to regularize an incomplete forward model and prevent the mis-estimation
 
sary to regularize an incomplete forward model and prevent the mis-estimation
Line 59: Line 59:
  
 
We present a variational approach for Atlas-based Susceptibility Mapping
 
We present a variational approach for Atlas-based Susceptibility Mapping
(ASM) that performs simultaneous susceptibility estimation and bias�field re-
+
(ASM) that performs simultaneous susceptibility estimation and biasfield re-
 
moval using the Laplacian operator and a tissue-air susceptibility atlas. In [7,
 
moval using the Laplacian operator and a tissue-air susceptibility atlas. In [7,
8, 6] it was shown that applying the Laplacian to the observed �field eliminates
+
8, 6] it was shown that applying the Laplacian to the observed field eliminates
 
non-local biasfields. Since MR systems contain active shims that produce bias-
 
non-local biasfields. Since MR systems contain active shims that produce bias-
�elds based on a spherical harmonic expansion [9], which is a solution to the
+
fields based on a spherical harmonic expansion [9], which is a solution to the
Laplace equation, the eff�ects of any mis-set shims and remote susceptibility dis-
+
Laplace equation, the effects of any mis-set shims and remote susceptibility dis-
tributions (ie. the neck/chest) are e�ffectively eliminated by taking the Laplacian
+
tributions (ie. the neck/chest) are effectively eliminated by taking the Laplacian
of the observed magnetic fi�eld. Since eigenfunction susceptibility distributions in the brain cannot be distin-
+
of the observed magnetic field. Since eigenfunction susceptibility distributions in the brain cannot be distin-
guished from low frequency bias�fields using phase information alone, additional
+
guished from low frequency biasfields using phase information alone, additional
 
modeling in the form of priors or atlases is needed to resolve this ambiguity.
 
modeling in the form of priors or atlases is needed to resolve this ambiguity.
 
In this method, large deviations from the susceptibility atlas are penalized,
 
In this method, large deviations from the susceptibility atlas are penalized,
Line 73: Line 73:
 
tissue-air boundaries where the Laplacian may not be sufficient to eliminate the
 
tissue-air boundaries where the Laplacian may not be sufficient to eliminate the
 
contribution of non-local sources and substantial signal loss corrupts the ob-
 
contribution of non-local sources and substantial signal loss corrupts the ob-
served �field. Agreement of predicted and observed �fields
+
served field. Agreement of predicted and observed fields
 
within the brain is also enforced, but deviations in estimated susceptibility values outside the
 
within the brain is also enforced, but deviations in estimated susceptibility values outside the
 
brain are not penalized, allowing values at the bound-
 
brain are not penalized, allowing values at the bound-
ary to vary from the atlas-based prior to account for unmodeled external �field
+
ary to vary from the atlas-based prior to account for unmodeled external field
 
sources (ie. shims).
 
sources (ie. shims).

Revision as of 20:22, 24 March 2011

Home < Projects:QuantitativeSusceptibilityMapping

Quantifying magnetic susceptibility in the brain from the phase of the MR signal provides a non-invasive means for measuring the accumulation of iron believed to occur with aging and neurodegenerative disease. Phase observations from local susceptibility distributions, however, are corrupted by external biasfields, which may be identical to the sources of interest. Furthermore, limited observations of the phase makes the inversion ill-posed. We describe a variational approach to susceptibility estimation that incorporates a tissue-air atlas to resolve ambiguity in the forward model, while eliminating additional biasfields through application of the Laplacian. Results show qualitative improvement over two methods commonly used to infer underlying susceptibility values, and quantitative susceptibility estimates show better correlation with postmortem iron concentrations than competing methods.

Description

There is increasing evidence that excessive iron deposition in specific regions of the brain is associated with neurodegenerative disorders such as Alzheimer's and Parkinson's disease [1]. The role of iron in the pathogenesis of these diseases remains unknown and is difficult to determine without a non-invasive method to quantify its concentration in-vivo. Since iron is a ferromagnetic substance, changes in iron concentration result in local changes in the magnetic suscepti- bility of tissue. In magnetic resonance imaging (MRI) experiments, differences in magnetic susceptibility cause perturbations in the local magnetic field, which can be computed from the phase of the MR signal. The advent of high field MRI has made it possible for subtle differences in susceptibility to measurably affect the phase of the MR signal. This has raised the possibility that estimating magnetic susceptibility from high field phase data may provide a new means for quantifying iron concentration in-vivo.

The field perturbations caused by magnetic susceptibility differences can be modeled as the convolution of a dipole-like kernel with the spatial susceptibility distribution. In the Fourier domain, the kernel exhibits zeros at the magic angle, preventing direct inversion of the fieldmap [3]. Critically, limited observations of the field make the problem ill-posed. The observed data is also corrupted by con- founding biasfields (ie. those from tissue-air interfaces, mis-set shims, and other non-local sources). Eliminating these fields is critical for accurate susceptibility estimation since they corrupt the phase contributions from local susceptibility sources.

In general, methods that rely heavily on agreement between observed and predicted field values computed using kernel-based forward models [2, 3, 6] are in- herently limited since they cannot distinguish between low frequency biasfields and susceptibility distributions that are eigenfunctions of the model. Examples of such distributions include constant, linear, and quadratic functions of suscep- tibility along the main field (ie. 'z') direction. Applying the forward model to these distributions results in predicted fields that are proportional to the local susceptibility sources, but also identical in form to non-local biasfields (ie. those produced by a z-shim). Therefore, removing all low frequency fields prior to susceptibility estimation will eliminate the biasfield as well as fields due to the sources of interest, potentially preventing accurate calculation of the underlying susceptibility values. In contrast, inadequate removal of the biasfield may result in the estimation of artifactual susceptibility eigenfunctions in areas where the biasfield is strong, such as regions adjacent to tissue-air interfaces. This suggests that additional information such as boundary conditions or priors may be neces- sary to regularize an incomplete forward model and prevent the mis-estimation of low frequency biasfields.

We present a variational approach for Atlas-based Susceptibility Mapping (ASM) that performs simultaneous susceptibility estimation and biasfield re- moval using the Laplacian operator and a tissue-air susceptibility atlas. In [7, 8, 6] it was shown that applying the Laplacian to the observed field eliminates non-local biasfields. Since MR systems contain active shims that produce bias- fields based on a spherical harmonic expansion [9], which is a solution to the Laplace equation, the effects of any mis-set shims and remote susceptibility dis- tributions (ie. the neck/chest) are effectively eliminated by taking the Laplacian of the observed magnetic field. Since eigenfunction susceptibility distributions in the brain cannot be distin- guished from low frequency biasfields using phase information alone, additional modeling in the form of priors or atlases is needed to resolve this ambiguity. In this method, large deviations from the susceptibility atlas are penalized, discouraging the estimation of artifactual susceptibility values in regions near tissue-air boundaries where the Laplacian may not be sufficient to eliminate the contribution of non-local sources and substantial signal loss corrupts the ob- served field. Agreement of predicted and observed fields within the brain is also enforced, but deviations in estimated susceptibility values outside the brain are not penalized, allowing values at the bound- ary to vary from the atlas-based prior to account for unmodeled external field sources (ie. shims).