Experiments on a New Inter- Subject Registration Method John Ashburner 2007
Experiments on a New Inter-Subject Registration Method
John Ashburner
2007
Abstract• The objective of this work was to devise a more precise method of inter-
subject brain image registration than those currently available in the SPM software. This involved a model with many more degrees of freedom, but which still enforces a one-to-one mapping. Speed considerations were also important. The result is an approach that models each warp by single velocity field. These are converted to deformations by a scaling and squaring procedure, and the inverses can be generated in a similar way. Registration is via a Levenberg-Marquardt optimization strategy, which uses a full multi-grid algorithm to rapidly solve the necessary equations.
• The method has been used for warping images of 471 subjects. This involved simultaneously matching grey matter with a grey matter template, and white matter with a white matter template. After every few iterations, the templates were re-generated from the means of the warped individual images. Evaluations involved applying pattern recognition procedures to the resulting deformations, in order to assess how well information such as the ages and sexes of the subjects could be predicted from the encoded deformations. A slight improvement in prediction accuracy was obtained when compared to a similar procedure using a small deformation model.
Overview
• Motivation– Dimensionality– Inverse-consistency
• Principles
• Geeky stuff
• Example
• Validation
• Future directions
Motivation
• More precise inter-subject alignment– Improved fMRI data analysis
• Better group analysis• More accurate localization
– Improve computational anatomy• More easily interpreted VBM• Better parameterization of brain shapes
– Other applications• Tissue segmentation• Structure labeling
Image Registration
• Figure out how to warp one image to match another
• Normally, all subjects’ scans are matched with a common template
Current SPM approach
• Only about 1000 parameters.– Unable model detailed
deformations
A simple 2D example
Individual brain
Warped Individual
Reference
Residual Differences
Individual brain
Warped Individual
Expansion and
contraction
• Relative volumes encoded by Jacobian determinants of deformation
Tissue volume comparisons
Warped grey matter
Jacobian determinants
Absolute grey matter
volumes
A one-to-one mapping
• Many models simply add a smooth displacement to an identity transform– One-to-one mapping not enforced
• Inverses approximately obtained by subtracting the displacement– Not a real inverse Small deformation
approximation
Overview
• Motivation
• Principles
• Geeky stuff
• Example
• Validation
• Future directions
Principles
DiffeomorphicAnatomicalRegistrationThroughExponentiatedLie Algebra
Deformations parameterized by a single flow field, which is considered to be constant in time.
DARTEL
• Parameterizing the deformation
• φ(0)(x) = x
• φ(1)(x) = ∫ u(φ(t)(x))dt• u is a flow field to be estimated
t=0
1
Euler integration
• The differential equation is
dφ(x)/dt = u(φ(t)(x))• By Euler integration
φ(t+h) = φ(t) + hu(φ(t))• Equivalent to
φ(t+h) = (x + hu) o φ(t)
Flow Field
For (e.g) 8 time steps
Simple integration• φ(1/8) = x + u/8• φ(2/8) = φ(1/8) o φ(1/8) • φ(3/8) = φ(1/8) o φ(2/8) • φ(4/8) = φ(1/8) o φ(3/8) • φ(5/8) = φ(1/8) o φ(4/8) • φ(6/8) = φ(1/8) o φ(5/8) • φ(7/8) = φ(1/8) o φ(6/8) • φ(8/8) = φ(1/8) o φ(7/8)
7 compositions
Scaling and squaring• φ(1/8) = x + u/8• φ(2/8) = φ(1/8) o φ(1/8)
• φ(4/8) = φ(2/8) o φ(2/8)
• φ(8/8) = φ(4/8) o φ(4/8)
3 compositions
• Similar procedure used for the inverse.Starts withφ(-1/8) = x - u/8
Scaling and squaring example
DARTEL
Jacobian determinants remain positive
Overview
• Motivation
• Principles
• Geeky stuff– Feel free to sleep
• Example
• Validation
• Future directions
Registration objective function
• Simultaneously minimize the sum of – Likelihood component
• From the sum of squares difference
• ½∑i(g(xi) – f(φ(1)(xi)))2
• φ(1) parameterized by u
– Prior component• A measure of deformation roughness
• ½uTHu
Regularization model
• DARTEL has three different models for H– Membrane energy– Linear elasticity– Bending energy
• H is very sparse
An example H for 2D registration of 6x6 images (linear elasticity)
Regularization models
Optimisation
• Uses Levenberg-Marquardt– Requires a matrix solution to a very large set
of equations at each iteration
u(k+1) = u(k) - (H+A)-1 b
– b are the first derivatives of objective function– A is a sparse matrix of second derivatives– Computed efficiently, making use of scaling
and squaring
Relaxation
• To solve Mx = cSplit M into E and F, where
• E is easy to invert• F is more difficult
• Sometimes: x(k+1) = E-1(c – F x(k))• Otherwise: x(k+1) = x(k) + (E+sI)-1(c – M x(k))
• Gauss-Siedel when done in place.• Jacobi’s method if not
• Fits high frequencies quickly, but low frequencies slowly
H+A = E+F
Highest resolution
Lowest resolution
Full Multi-Grid
Overview
• Motivation
• Principles
• Geeky stuff
• Example– Simultaneous registration of GM & WM– Tissue probability map creation
• Validation
• Future directions
Simultaneous registration of GM to GM and WM to WM
Grey matter
White matter
Grey matter
White matter
Grey matter
White matter
Grey matter
White matter
Grey matter
White matterTemplate
Subject 1
Subject 2
Subject 3
Subject 4
TemplateInitial
Average
After a few iterations
Final template
Iteratively generated from 471 subjects
Began with rigidly aligned tissue probability maps
Used an inverse consistent formulation
Grey matter average of 452 subjects – affine
Grey matter average of 471 subjects
White matter average of 471 subjects
Initial GM images
Warped GM images
Overview
• Motivation
• Principles
• Geeky stuff
• Example
• Validation– Sex classification– Age regression
• Future directions
Validation
• There is no “ground truth”• Looked at predictive accuracy
– Can information encoded by the method make predictions?
• Registration method blind to the predicted information• Could have used an overlap of fMRI results
– Chose to see whether ages and sexes of subjects could be predicted from the deformations
• Comparison with small deformation model
Training and Classifying
ControlTraining Data
PatientTraining Data
?
?
??
Classifying
Controls
Patients
?
?
??
y=f(aTx+b)
Support Vector Classifier
Support Vector Classifier (SVC)
SupportVector
SupportVector
Support
Vector
a is a weighted linear combination of the support vectors
Some Equations
• Linear classification is by y = f(aTx + b)– where a is a weighting vector, x is the test data, b is an
offset, and f(.) is a thresholding operation
• a is a linear combination of SVs a = i wi xi
• So y = f(i wi xiTx + b)
Going Nonlinear
• Nonlinear classification is by
y = f(i wi (xi,x))– where (xi,x) is some function of xi and x.
• e.g. RBF classification (xi,x) = exp(-||xi-x||2/(22))
• Requires a matrix of distance measures (metrics) between each pair of images.
Nonlinear SVC
Cross-validation
• Methods must be able to generalise to new data• Various control parameters
– More complexity -> better separation of training data– Less complexity -> better generalisation
• Optimal control parameters determined by cross-validation– Test with data not used for training– Use control parameters that work best for these data
Two-fold Cross-validation
Use half the data for training.
and the other half for testing.
Two-fold Cross-validation
Then swap around the training and test data.
Leave One Out Cross-validation
Use all data except one point for training.
The one that was left out is used for testing.
Leave One Out Cross-validation
Then leave another point out.
And so on...
Support-vector classification
• Guess sexes of 471 subjects from brain shapes – 207 Females / 264 Males
• Use a random sample of 400 for training.
• Test on the remaining 71.
• Repeat 50 times.
Sex classification results
• Small Deformation– Linear classifier
• 87.0% correct• Kappa = 0.736
– RBF classifier• 87.1% correct• Kappa = 0.737
• DARTEL– Linear classifier
• 87.7% correct• Kappa = 0.749
– RBF classifier• 87.6% correct• Kappa = 0.748
An unconvincing improvement
Regression
23
26
30
29
18
32
40
Relevance-vector regression
• A Bayesian method, related to SVMs– Developed by Mike Tipping
• Guess ages of 471 subjects from brain shapes.
• Use a random sample of 400 for training.
• Test on the remaining 71.
• Repeat 50 times.
Age regression results
• Small deformation– Linear regression
• RMS error = 7.55• Correlation = 0.836
– RBF regression• RMS error = 6.68• Correlation = 0.856
• DARTEL– Linear regression
• RMS error = 7.90• Correlation = 0.813
– RBF regression• RMS error = 6.50• Correlation = 0.867
An unconvincing improvement(slightly worse for linear regression)
Overview
• Motivation
• Principles
• Geeky stuff
• Example
• Validation
• Future directions
Future directions
• Compare with variable velocity methods– Beg’s LDDMM algorithm
• Classification/regression from “initial momentum”
• Combine with SPM5 segmentation model– Similar to Emiliano D’Agostino’s method
• Develop a proper EM framework for generating tissue probability maps
u
Hu
“Initial momentum”
Variable velocity framework
“Initial momentum”
Variable velocity framework
Thank you