Spatial Preprocessing Ged Ridgway University College London Thanks to: John Ashburner, Klaas Enno Stephan, and the FIL Methods Group. Meike Grol, Chloe Hutton, Jesper Andersson, Floris de Lange, Miranda van Turennout, Lennart Verhagen, … Terminology of fMRI Structural (T1) images: - high resolution - to distinguish different types of tissue Functional (T2*) images: - lower spatial resolution - to relate changes in BOLD signal to an experimental manipulation Time series: A large number of images that are acquired in temporal order at a specific rate t Terminology of fMRI subjects sessions runs single run volume slices TR = repetition time time required to scan one volume voxel Terminology of fMRI Scan Volume: Field of View (FOV), e.g. 192 mm Axial slices Slice thickness e.g., 3 mm Matrix Size e.g., 64 x 64 In-plane resolution 192 mm / 64 = 3 mm 3 mm 3 mm 3 mm Voxel Size (volumetric pixel) fMRI Time-series Movie Overview of SPM Analysis fMRI time-series Motion Correction Statistical Parametric Map Design matrix Smoothing Spatial Normalisation Anatomical Reference General Linear Model Parameter Estimates Preprocessing in SPM • Realignment – With non-linear unwarping for EPI fMRI • • • • • Slice-time correction Coregistration Normalisation Segmentation Smoothing The many guises of affine registration • Manual reorientation • Rigid intra-modal realignment – Motion correction of fMRI time-series – Within-subject longitudinal registration of serial sMRI • Rigid inter-modal coregistration – Aligning structural and (mean) functional images – Multimodal structural registration, e.g. T1-T2 • Affine inter-subject registration – First stage of non-linear spatial normalisation – Approximate alignment of tissue probability maps Other types of registration in SPM • Non-linear spatial normalisation – Registering different subjects to a standard template • Unified segmentation and normalisation – Warping standard-space tissue probability maps to a particular subject (can normalise using the inverse) • High-dimensional warping – Modelling small longitudinal deformations (e.g. AD) • DARTEL – Smooth large-deformation warps using flows – Normalisation to group’s average shape template Manual reorientation Image “headers” contain information that lets us map from voxel indices to “world” coordinates in mm Modifying this mapping lets us reorient (and realign or coregister) the image(s) Manual reorientation Manual reorientation Interpolation Nearest (Bi)linear Neighbour (Bi)linear Sinc Interpolation • Applying the transformation parameters, and resampling the data onto the same grid of voxels as the target image (usually) – AKA reslicing, regridding, transformation, and writing (as in normalise - write) • Nearest neighbour gives the new voxel the value of the closest corresponding voxel in the source • Linear interpolation uses information from all immediate neighbours (2 in 1D, 4 in 2D, 8 in 3D) • NN and linear interp. correspond to zeroth and first order B-spline interpolation, higher orders use more information in the hope of improving results – (Sinc interpolation is an alternative to B-spline) Manual reorientation – Reslicing Reoriented Resliced (1x1x3 mm voxel size) (to 2 mm cubic) Quantifying image alignment • Registration intuitively relies on the concept of aligning images to increase their similarity – This needs to be mathematically formalised – We need practical way(s) of measuring similarity • Using interpolation we can find the intensity at equivalent voxels – (equivalent according to the current transformation parameter estimates) Voxel similarity measures I ref (i, j , k ) I (i 1, j , k ) ref I ref (i, j 1, k ) Pairs of voxel intensities • Mean-squared difference • Correlation coefficient • Joint histogram measures I src (i, j , k ) I (i 1, j , k ) src I src (i, j 1, k ) Intra-modal similarity measures • Mean squared error (minimise) – AKA sum-squared error, RMS error, etc. (monotonically related). AKA “least squares” – Assumes simple relationship between intensities – Optimal (only) if differences are i.i.d. Gaussian – Okay for fMRI realignment or sMRI-sMRI coreg • Correlation-coefficient (maximise) – Slightly more general, e.g. T1-T1 inter-scanner – Invariant under affine transformation of intensities – AKA Normalised Cross-Correlation, Zero-NCC Automatic image registration • Quantifying the quality of the alignment with a measure of image similarity allows computational estimation of transformation parameters • This is the basis of both realignment and coregistration in SPM – Allowing more complex geometric transformations or warps leads to more flexible spatial normalisation • Let’s look at an example of realignment next – We will return to coregistration later... Option of second pass, reregistering to mean from first pass, improves results if first image is not perfectly representative of all others How much motion is significant? Edge: DS(x) ~ 70-90% 0.05 pixel shift 3 - 5% DS (187 mm) Inside: DS(x) ~ 10-20% 0.25 pixel shift (~1 mm) Slide from Kent Kiehl’s 2005 (USA) Course Workshop 3 - 5% DS Motion in fMRI • Even small head movements can be a major problem: – Movement artefacts add up to the residual variance and reduce sensitivity. – Data may get completely lost if sudden movements occur during a single volume – Movements may be correlated with the task performed Minimising movements is one of the most important factors for ensuring good data quality fMRI motion prevention 1. Constrain the volunteer’s head 2. Instruct him/her explicitly to remain as calm as possible, not to talk between sessions, and swallow as little as possible 3. Do not scan for too long – everyone will move after while! fMRI motion correction • Each volume in the time-series is realigned – Using rigid-body registration • Assumes that brain shape doesn’t change – Least squares cost function • Realigned images must be resliced for analysis – Not necessary if they will be normalised anyway • An example of severe motion… Residual errors in realigned fMRI Even after realignment, as much as 90% of the variance can be accounted for by effects of movement This can be caused by e.g.: 1. Movement between and within slice acquisition 2. Interpolation artefacts due to resampling 3. Non-linear distortions and drop-out due to inhomogeneity of the magnetic field Incorporate movement parameters as confounds in the statistical model Movement-by-distortion interaction • Subject disrupts B0 field, rendering it inhomogeneous – distortions occur along phase-encode direction • Subject moves during EPI time series – Distortions vary with subject orientation – Shape varies Co-registration example • Rigid intra-modality inter-subject registration • Purely to illustrate concepts • Coreg more commonly used to register a structural image to a mean functional one – Allows more accurate anatomical localisation – Assists spatial normalisation (see later) – Examples in SPM8 Manual, Chapters 28, 29 Reference AKA template AKA fixed image (single_subj_T1) Source AKA template (!) AKA moving image (auditory/sM00223) spm_defaults; global defaults opts = defaults.coreg.estimate opts = cost_fun: 'nmi' sep: [4 2]iteration 1... tol: [1x123.72 double] 0 15.95 fwhm: [7 7]3.72 3.72 % intra-modal 15.95 opts.cost_fun = 'ncc'; 3.72 15.95 15.95 x = spm_coreg(ref, 3.72 src, opts) % … 3.72 15.95 iteration 2... x = ... -0.02 -0.01 2.52 20.86 28.79 -0.20 iteration 8... M = spm_matrix(x) 2.447 20.97 M = 2.447 20.97 0.999 -0.011 2.447 -0.016 20.97 2.522 0.008 0.981 2.447 -0.195 20.97 20.857 0.018 0.195 2.447 0.981 20.97 28.790 0 0 2.447 0 20.97 1.000 det(M) ans = 1.000 0 0 19.61 19.61 19.61 19.61 0 0 0 -0.1837 -0.1837 -0.1837 0 0 0 0 -0.00151 -0.00151 28.73 28.73 28.74 28.74 28.74 28.74 -0.1905 -0.1905 -0.1905 -0.1905 -0.1905 -0.1905 -0.003633 -0.003633 -0.003633 -0.003633 -0.003633 -0.003633 Svx2Rvx = inv(ref.mat)*inv(M)*src.mat 0 0 0 0 0 -0.00063 -0.01026 -0.01026 -0.01026 -0.01026 -0.01026 -0.01031 | | | | | | -0.50731 -0.55893 -0.62815 -0.6434 -0.64341 -0.64341 | | | | | | -0.71486 -0.71486 -0.71487 -0.71487 -0.71487 -0.71487 Inter-modal similarity measures • Often derived from joint and marginal entropies – Entropies via probabilities from histograms • Mutual Information (maximise) – MI= ab P(a,b) log2 [P(a,b)/( P(a) P(b) )] – Related to entropy: MI = H(a) + H(b) - H(a,b) • H(a) = -a P(a) log2P(a) • H(a,b) = -a P(a,b) log2P(a,b) • Normalised MI = (H(a) + H(b)) / H(a,b) Joint and marginal histograms Joint histogram based registration Initially registered T1 and T2 templates After deliberate misregistration (10mm relative x-translation) Joint histogram sharpness correlates with image alignment Mutual information and related measures attempt to quantify this Overview of SPM Analysis fMRI time-series Motion Correction Statistical Parametric Map Design matrix Smoothing Spatial Normalisation Anatomical Reference General Linear Model Parameter Estimates Spatial Normalisation Spatial Normalisation - Reasons • Inter-subject averaging – Increase sensitivity with more subjects • Fixed-effects analysis – Extrapolate findings to the population as a whole • Mixed-effects analysis • Make results from different studies comparable by aligning them to standard space – e.g. The T&T convention, using the MNI template Standard spaces The Talairach Atlas The MNI/ICBM AVG152 Template The MNI template follows the convention of T&T, but doesn’t match the particular brain Recommended reading: http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach Coordinate system sense • Analyze™ files are stored in a left-handed system • Talairach space has the opposite (right-handed) sense • Mapping between them requires a reflection or “flip” – Affine transform with a negative determinant z x z y y y z x x Rotated example Spatial normalisation – Issues We want to warp the images to match functionally homologous regions from different subjects However… 1. No exact match between structure and function 2. Different brains – different structural topology 3. Computational problems (local minima, etc.) Compromise by correcting gross differences followed by smoothing of normalised images Spatial Normalisation – Procedure • Start with a 12 DF affine registration – 3 translations, 3 rotations 3 zooms, 3 shears – Fits overall shape and size • Refine the registration with non-linear deformations • Algorithm simultaneously minimises – Mean-squared difference (Gaussian likelihood) – Squared distance between parameters and their expected values (regularisation with Gaussian prior) Spatial Normalisation – Warping Deformations are modelled with a linear combination of non-linear basis functions Spatial Normalisation – DCT basis The lowest frequencies of a 3D discrete cosine transform (DCT) provide a smooth basis spm_dctmtx(5,5) ans = 0.447 0.602 0.512 0.447 0.372 -0.195 0.447 0.000 -0.633 0.447 -0.372 -0.195 0.447 -0.601 0.512 plot(spm_dctmtx(50, 5)) 0.372 -0.602 -0.000 0.602 -0.372 0.195 -0.512 0.633 -0.512 0.195 % Note, pinv(x)=x’, projection P=x*x’ P{n} = x(:,1:n)*x(:,1:n)’ P{N} == eye(N) P{n<N} projects to smoother approx. Spatial Normalisation – Templates and masks A wider range of contrasts can be registered to a linear combination of template images. PET T2 T1 Transm EPI PD PET Spatial normalisation can be weighted so that non-brain voxels do not influence the result. T1 PD More specific weighting masks can be used to improve normalisation of lesioned brains. Spatial Normalisation – Results Affine registration Non-linear registration Optimisation • Find the “best” parameters according to an “objective function” (minimised or maximised) • Objective functions can often be related to a probabilistic model, e.g. Maximum Likelihood Objective function Global optimum (most probable) Local optimum Value of parameter Local optimum Optimisation – poor local optima Optimisation – regularisation • The “best” parameters according to the objective function may not be realistic • In addition to similarity, regularisation terms or constraints are often needed to ensure a reasonable solution is found – Also helps avoid poor local optima – These can be considered as priors in a Bayesian framework, e.g. converting ML to MAP: • log(posterior) = log(likelihood) + log(prior) + c Spatial Normalisation – Overfitting Without regularisation, the non-linear normalisation can introduce unnecessary deformation Template image Non-linear registration using regularisation. (2 = 302.7) Affine registration. (2 = 472.1) Non-linear registration without regularisation. (2 = 287.3) Normalising fMRI time-series • Two main options for fMRI normalisation: 1. Structural normalisation (unified segmentation) – Rigidly coregister structural MRI to mean functional – Normalise structural and apply warp (write) to fMRI 2. EPI normalisation – Normalise mean functional to EPI template – Apply warp to fMRI time-series – Optional coregistration and norm-write of structural for visualisation purposes Normalising structural to T1 template Estimate normalization parameters (T1 -> T1 template) T1 image Apply the normalization parameters to the T1 image EPI time series And apply the normalization parameters to the (coregistered) functional images Normalising time series to EPI template Calculate mean of EPI time series Estimate normalization parameters (EPI -> EPI template) EPI time series Apply the normalization parameters to all EPI images T1 image Apply the normalization parameters to the (coregistered) T1 image Custom EPI template Custom EPI template (note different coords) Original EPI image Normalised EPI image Smoothing • Why would we deliberately blur the data? – Suppress noise and effects due to differences in anatomy by averaging over neighbouring voxels – Achieve better spatial overlap, enhanced sensitivity, and greater validity of statistical assumptions – Reduce the effective number of multiple comparisons when correcting results using Random Field Theory • How is it implemented? – Convolution with a 3D Gaussian kernel, of specified Full-width at half-maximum (FWHM) in mm Example of Gaussian smoothing in one-dimension The Gaussian kernel is separable we can smooth 2D data with two 1D convolutions. Generalisation to 3D is simple and efficient A 2D Gaussian Kernel Smoothing – a link to ROI analysis Each voxel after smoothing effectively represents a weighted average over its local region of interest (ROI) Before convolution Convolved with a circle Gaussian convolution Preprocessing in SPM • Realignment – With non-linear unwarping for EPI fMRI • • • • • Slice-time correction Coregistration SPM8’s unified tissue segmentation Normalisation and spatial normalisation procedure will be covered in the VBM lecture Segmentation Smoothing