-1-! Alignment and Atlases � (volume registration and Talairach transformation)! -2-! What does aligning mean and why do we want to do it? •  Alignment means to bring two objects into the same space so that each location within one object corresponds to the same location in the other! •  Why?! motion correction across time!   align EPI to anatomical data or vice versa – to assign a location with a   functional result! compare data from longitudinal studies!   compare data from different scanners, sites!   compare results with a standard template or atlas for standardized locations   and structures! -3-! Alignment goals and tools in AFNI •  EPI data across time in a single run or across runs to a base image! 3dvolreg – motion correction (rigid)!   •  align data to template! 3dWarpDrive, @auto_tlrc – align similar volumes (affine) even across   subjects! 3dQwarp, auto_warp.py – align similar volumes nonlinearly to template!   •  align images across modalities – EPI to anat! 3dAllineate – align different or similar volumes!   align_epi_anat.py – general alignment script to align EPI with anatomical   data! •  Include motion correction, alignment of EPI to anatomical in fMRI processing pipeline script ! afni_proc.py!   •  Correct for motion between two volumes by aligning in two dimensions using corresponding slices! @2dwarper.Allin – non-linear alignment of slices!   @2dwarper, 2dimreg limit alignment to specific plane!  -4-! Alignment tools in AFNI (continued) •  align partial data to roughly the right part of the brain! Nudge plug-in - visually align two volumes!   •  rotate by known amount between volumes! 3drotate – moves (shifts and rotates) volumes!   3dWarp – make oblique, deoblique to match another dataset!   •  Put centers of data from outside sources in roughly the same space! @Align_Centers, 3dCM – put centers or centers of mass of dataset in   same place! •  align specific regions across subjects! 3dTagalign, tagset plugin – place and align volumes using corresponding   fiducial marker points! •  align one jpeg image to another! imreg – align two 2D images!  -5-! Image and Volume Registration with AFNI •  Goal: bring images collected with different methods and at different times into spatial alignment! •  Facilitates comparison of data on a voxel-by-voxel basis! Functional time series data will be less contaminated by artifacts due to subject   movement! Can compare results across scanning sessions once images are properly registered!   Can put volumes in standard space such as the stereotaxic Talairach-Tournoux   coordinates! •  Most (all?) image registration methods now in use do pair-wise alignment:!   Given a base image J(x) and target (or source) image I(x), find a geometrical transformation T[x] so that I(T[x])≈J(x)! T[x] will depend on some parameters!   ➥  Goal is to find the parameters that make the transformed I a ‘best fit’ to J! To register an entire time series, each volume I (x) is aligned to J(x) with its own   n transformation T [x], for n=0, 1, …! n ➥  Result is time series I (T [x]) for n=0, 1, …! n n ➥  User must choose base image J(x)! -6-! •  Most image registration methods make 3 algorithmic choices:! How to measure mismatch E (for error) between I(T[x]) and J(x)?!   ➥  Or … How to measure goodness of fit between I(T[x]) and J(x)?! ➭  E(parameters) ≡ –Goodness(parameters)! How to adjust parameters of T[x] to minimize E?!   How to interpolate I(T[x]) to the J(x) grid?!   ➥  So we can compare voxel intensities directly! •  The input volume is transformed by the optimal T[x] and a record of the transform is kept in the header of the output. ! •  Finding the transform to minimize E is the bulk of the registration work. Applying the transform is easy and is done on the fly in many cases.! •  If data starts off far from each other, may add a coarse pass (twopass) step! guess a lot among all the parameters (rotations, shifts, ...), measure cost!   best guesses, tweak the parameters (optimize) and measure again!   Now, applications of alignment…! -7-! Preprocess – mask, weight Alignment process - overview Initial interpolation, cost •  Preprocess – mask data, weight data! Lots of •  If far off, take some random guesses (-twopass)! random twopass? •  Optimize parameters on initial or best sets guesses – follow n best (6,12,39,1000's)! Use new parameters to transform input! o  Tweak Interpolate onto base data's grid!   parameters Measure alignment error with cost functional! o  Less than minimum error - finished!   n o   Better - keep adjusting with same Transform & i t direction! Interpolate a z i Worse – try other parameters ! m   •  Create final output by interpolating onto output ti Measure cost p grid! O save datasets, transform parameters! o  Cost < tolerance? Save final data -8-! Within Modality Registration -9-! •  AFNI program 3dvolreg is for aligning 3D volumes by rigid movements! T[x] has 6 parameters:!   ➥  Shifts along x-, y-, and z-axes; Rotations about x-, y-, and z-axes! Generically useful for intra- and inter-session alignment!   Motions that occur within a single TR (2-3 s) cannot be corrected this way, since   method assumes rigid movement of the entire volume! •  AFNI program 3dWarpDrive is for aligning 3D volumes by affine transformations! T[x] has up to 12 parameters:!   ➥  Same as 3dvolreg plus 3 Scales and 3 Shears along x-, y-, and z-axes! Generically useful for intra- and inter-session alignment!   Generically useful for intra- and inter-subject alignment!   •  AFNI program 2dImReg is for aligning 2D slices! T[x] has 3 parameters for each slice in volume:!   ➥  Shift along x-, y-axes; Rotation about z-axis! ➥  No out of slice plane shifts or rotations!! Useful for sagittal EPI scans where dominant subject movement is ‘nodding’ motion   that may be faster than TR! It is possible and sometimes even useful to run 2dImReg to clean up sagittal nodding   motion, followed by 3dvolreg to deal with out-of-slice motion! -10-!•  Intra-session registration example:! ! !3dvolreg -base 4 -heptic -zpad 4 !\! ! ! ! -prefix fred1_epi_vr !\! ! ! ! -1Dfile fred1_vr_dfile.1D \! ! ! ! fred1_epi+orig! Input dataset name!   -base 4 ⇒ Selects sub-brick #4 of dataset fred1_epi+orig as base image J(x)!   -heptic ⇒ Use 7th order polynomial interpolation!   -zpad 4 ⇒ Pad each target image, I(x), with layers of zero voxels 4 deep on each face prior to shift/rotation, then strip them off afterwards (before output)! ➥  Zero padding is particularly desirable for -Fourier interpolation! also good for large rotations, some data may get ‘lost’ if no zero padding! ➥    -prefix fred1_epi_vr ⇒ Save output dataset into a new dataset with the given prefix name (e.g., fred1_epi_vr+orig)!   -1Dfile fred1_vr_dfile.1D ⇒ Save estimated movement parameters into a 1D (i.e., text) file with the given name! Movement parameters can be plotted with the 1dplot command and used ➥  later....! Try this (in AFNI_data6/afni) :! !3dvolreg -base 3 -cubic -prefix epi_r1_vrt -1Dfile vr_dfile.1D epi_r1+orig! !1dplot -volreg vr_dfile.1D &