FIL SPM Course 2010 Spatial Preprocessing By John Ashburner & Ged Ridgway.
Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen...
-
Upload
julie-carson -
Category
Documents
-
view
217 -
download
4
Transcript of Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen...
![Page 1: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/1.jpg)
Preprocessing II:Between Subjects
John Ashburner
Wellcome Trust Centre for Neuroimaging,
12 Queen Square, London, UK.
![Page 2: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/2.jpg)
Pre-processing overviewfMRI time-series
Motion Correct
Coregister
1000
34333231
24232221
14131211
mmmm
mmmm
mmmm
Deformation
Estimate Spatial Norm
Spatially normalised
Smooth
Smoothed
Statistics or whatever
TemplateAnatomical MRI
![Page 3: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/3.jpg)
Alternative pipelinefMRI time-series
Motion Correct
Deformation
Estimate Spatial Norm
Spatially normalised
Smooth
Smoothed
Statistics or whatever
Template
![Page 4: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/4.jpg)
Contents* Normalise/Segment
Use segmentation routine for spatial normalisation* Gaussian mixture model* Intensity non-uniformity correction* Deformed tissue probability maps
* Dartel* Smoothing
![Page 5: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/5.jpg)
Spatial normalisation* Brains of different subjects vary in shape and size.* Need to bring them all into a common anatomical space.
* Examine homologous regions across subjects* Improve anatomical specificity* Improve sensitivity
* Report findings in a common anatomical space (eg MNI space)
* In SPM, alignment is achieved by matching grey matter with grey matter and white matter with white matter.* Need to segment.
![Page 6: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/6.jpg)
Normalise/Segment
* This is the same algorithm as for tissue segmentation.
* Combines:* Mixture of Gaussians (MOG)* Bias Correction Component* Warping (Non-linear
Registration) Component
![Page 7: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/7.jpg)
Spatial normalisation
* Default spatial normalisation in SPM12 estimates nonlinear warps that match tissue probability maps to the individual image.
* Spatial normalisation achieved using the inverse of this transform.
![Page 8: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/8.jpg)
Segmentation
* Segmentation in SPM12 also estimates a spatial transformation that can be used for spatially normalising images.
* It uses a generative model, which involves:* Mixture of Gaussians (MOG)* Warping (Non-linear
Registration) Component* Bias Correction Component
![Page 9: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/9.jpg)
Image Intensity Distributions (T1-weighted MRI)
![Page 10: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/10.jpg)
Modelling tissue intensities
* Classification is based on a Mixture of Gaussians model (MOG), which represents the intensity probability density by a number of Gaussian distributions.
Image Intensity
Frequency
![Page 11: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/11.jpg)
Modelling deformations
![Page 12: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/12.jpg)
Modelling a bias field
* A bias field is modelled as a linear combination of basis functions.
Corrupted image Corrected imageBias Field
![Page 13: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/13.jpg)
Iterative optimisation scheme
Update tissue estimates
Update bias field estimatesUpdate deformation estimates
Converged?
Yes
No
![Page 14: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/14.jpg)
Evaluations of nonlinear
registration algorithms
![Page 15: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/15.jpg)
Old tissue probability maps
* Tissue probability maps (TPMs) are used instead of the proportion of voxels in each Gaussian as the prior.
ICBM Tissue Probabilistic Atlases. These tissue probability maps are kindly provided by the International Consortium for Brain
Mapping, John C. Mazziotta and Arthur W. Toga.
![Page 16: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/16.jpg)
Tissue probability maps in SPM12
Includes additional non-brain tissue classes (bone, and soft
tissue)
![Page 17: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/17.jpg)
![Page 18: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/18.jpg)
Contents* Normalise/Segment
* Dartel* Velocity field parameterisation* Objective function* Template creation* Examples
* Smooth
![Page 19: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/19.jpg)
Dartel image registration* Uses fast approximations
* Deformation integrated using scaling and squaring
* Uses Levenberg-Marquardt optimiser* Multi-grid matrix solver
* Matches GM with GM, WM with WM etc
* Diffeomorphic registration takes about 30 mins per image pair (121×145×121 images).
Grey matter template warped to
individual
Individual scan
![Page 20: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/20.jpg)
Dartel* Parameterising the deformation
* φ(0) = Identity* φ(1) = ∫ v(φ(t))dt* v is an estimated velocity field.
* Scaling and squaring is used to generate deformations.
t=0
1
![Page 21: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/21.jpg)
Scaling and squaring example
![Page 22: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/22.jpg)
Registration objective function* Simultaneously minimize the sum of:
* Matching Term* Drives the alignment of the images.* Multinomial assumption
* Regularisation term* A measure of deformation roughness* Keeps the warps spatially smooth.
* A balance between the two terms.
![Page 23: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/23.jpg)
Effect of different forms of regularisation
![Page 24: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/24.jpg)
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 matter
Template
Subject 1
Subject 2
Subject 3
Subject 4
![Page 25: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/25.jpg)
TemplateInitial Average
After a few iterations
Final template
Iteratively generated from 471
subjects
Began with rigidly aligned tissue
probability maps
![Page 26: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/26.jpg)
Grey matter average of 452
subjects – affine
![Page 27: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/27.jpg)
Grey matter average of 471
subjects
![Page 28: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/28.jpg)
Initial
GM images
![Page 29: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/29.jpg)
Aligned
GM images
![Page 30: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/30.jpg)
471 Subject Average
![Page 31: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/31.jpg)
471 Subject Average
![Page 32: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/32.jpg)
471 Subject Average
![Page 33: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/33.jpg)
Subject 1
![Page 34: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/34.jpg)
471 Subject Average
![Page 35: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/35.jpg)
Subject 2
![Page 36: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/36.jpg)
471 Subject Average
![Page 37: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/37.jpg)
Subject 3
![Page 38: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/38.jpg)
471 Subject Average
![Page 39: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/39.jpg)
Evaluations of nonlinear
registration algorithms
![Page 40: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/40.jpg)
Contents* Normalise/Segment* Dartel
* Smoothing* Compensating for inaccuracies in inter-subject alignment
![Page 41: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/41.jpg)
Before convolution Convolved with a circle Convolved with a Gaussian
Blurring is done by convolution.
Each voxel after smoothing effectively becomes the result of applying a weighted region of
interest (ROI).
Smooth
![Page 42: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/42.jpg)
Smooth
![Page 43: Preprocessing II: Between Subjects John Ashburner Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London, UK.](https://reader035.fdocuments.us/reader035/viewer/2022062715/56649d825503460f94a67e6c/html5/thumbnails/43.jpg)
References* Ashburner & Friston. Unified Segmentation.
NeuroImage 26:839-851 (2005).* Ashburner. A Fast Diffeomorphic Image Registration
Algorithm. NeuroImage 38:95-113 (2007).* Ashburner & Friston. Computing average shaped tissue
probability templates. NeuroImage 45(2): 333-341 (2009).
* Klein et al. Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration. NeuroImage 46(3):786-802 (2009).