Voxel-based morphometry

Download Report

Transcript Voxel-based morphometry

Voxel-based morphometry
The methods and the interpretation (SPM based)
Harma Meffert
Methodology meeting
14 april 2009
Outline
•
•
•
•
•
General preprocessing steps
Preprocessing
Comparison two recent tools
Data analysis
Discussion about ‘ISSUES’
General preprocessing steps …
VBM
General preprocessing steps
anatomical
scan
segmentation
normalisation
smoothing
VBM
Normalisation step; a closer look
1. Determine
parameters
VBM
Normalisation step; a closer look
1. Determine
parameters
2. Deform brain to
fit template
VBM
Normalisation step; a closer look
1. Determine
parameters
2. Deform brain to fit
template
3. Unmodulated
(concentration)
4. Modulated
(volumetric)
Unmodulated * Volume before warping / Volume after warping
Preprocessing …
Protocols and toolboxes
Overview ‘toolboxes’ and protocols
•
•
•
•
•
•
•
Standard VBM – SPM99 / SPM2
Optimised VBM – SPM99 / SPM2
VBM with unified segmentation – SPM5
VBM2 toolbox for SPM2
VBM5 toolbox for SPM5
Dartell
…
Standard VBM – SPM99 / SPM2
Normalisation
Segmentation
Gray matter
White matter
modulation
modulation
smoothing
smoothing
Analysis
Analysis
Mechelli et al. 2005
Optimised VBM – SPM99 / SPM2
Segmentation
Gray matter
White matter
Normalisation to
GM template
Normalisation to
WM template
Apply norm. par.
to raw image
Apply norm. par.
to raw image
modulation
modulation
smoothing
smoothing
Analysis
Analysis
Mechelli et al. 2005
VBM with unified segmentation – SPM5
Tissue classification, image registration and bias
correction within one model
Normalisation / segmentation
modulation
smoothing
Analysis
VBM5 toolbox in SPM5
MRF prior probability
Noise reduction with Markov Random Field
Summary: Segmentation and Normalisation
Options and considerations:
–
–
–
–
Normalisation before segmentation
Optimized order (norm  segm  norm)
Unified segmentation (SPM5)
Unified segmentation with the use of customized
priors (VBM5)
– Unified segmentation without the use of priors for
tissue classification (VBM5)
– Hidden Markov Random Field (VBM5)
– Center of mass as origin doesn’t work
Summary: Modulation
Options, considerations and questions
– Unmodulated ≈ ‘concentration’
– Modulated ≈ ‘volume’
– Modulation of …
• non-linear effects only
• affine and non-linear effects (no correction for
brain size afterwards)
– Smoothing
– Less smoothing in modulated images
Comparison two recent tools…
VBM5 vs SPM5
Data analysis …
Data-analysis: Considerations
• Corrections for multiple comparisons with
local maxima of the t statistic
• GLM with SPM, SnPM, machine learning
algorithms
• Global or localized inferences? Use of
covariates
• Non-stationary cluster extent correction
Voxel-based morphometry …
The Issues!
Issue 1: Unmodulated images…
•
•
•
•
Compatible with modulated images?
Just registration errors?
Very dependend on used toolbox?
Normalisation proces: Adding or removing
voxels… how does that happen?
Issue 2: Covariates
• If you modulate for both affine and nonlinear effects you do not have to correct for
global brain size….
• If global brain size is correlated with
‘treatment’ it is not a good covariate
because it will mask ‘treatment’ effects
Issue 3: What do the tissue labels mean
• If you add up probabilities in one voxel
across different tissue types they can be
>1
• Could you use white and gray maps to
determine the relative amount of gray for
example
Issue 4: How do you assess the quality of
segmentation
• VBM5 has the option to chack sample
homogeneity
• Furthermore it is visual inspection
Literature
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Ashburner, J. and K. J. Friston (2000). "Voxel-based morphometry--the methods." Neuroimage 11(6 Pt 1): 805-21.
Ashburner, J. and K. J. Friston (2001). "Why voxel-based morphometry should be used." Neuroimage 14(6): 123843.
Ashburner, J. and K. J. Friston (2005). "Unified segmentation." Neuroimage 26(3): 839-51.
Bookstein, F. L. (2001). ""Voxel-based morphometry" should not be used with imperfectly registered images."
Neuroimage 14(6): 1454-62.
Devlin, J. T. and R. A. Poldrack (2007). "In praise of tedious anatomy." Neuroimage 37(4): 1033-41; discussion
1050-8.
Good, C. D., I. S. Johnsrude, et al. (2001). "A voxel-based morphometric study of ageing in 465 normal adult
human brains." Neuroimage 14(1 Pt 1): 21-36.
Mechelli, A., C. J. Price, et al. (2005). "Voxel-based morphometry of the human brain: Methods and applications."
Current Medical Imaging Reviews 1(2): 105-113.
Ridgway, G. R., S. M. Henley, et al. (2008). "Ten simple rules for reporting voxel-based morphometry studies."
Neuroimage 40(4): 1429-35.
Ridgway, G. R., R. Omar, et al. (2009). "Issues with threshold masking in voxel-based morphometry of atrophied
brains." Neuroimage 44(1): 99-111.
NeuroImaging Center – Social Brain
lab:
1.
Prof. Dr. Christian Keysers
2.
Dr. Valeria Gazzola
3.
MSc. Jojanneke Bastiaansen
4.
Other members of the lab
Department of Psychiatry, UMCG
Prof. Dr. Hans den Boer
FPC Dr. S. van Mesdag
1.
Dr. Arnold Bartels
2.
Dr. Marinus Spreen
3.
Research department