2)    Realignment: Outside of SPM, known as ‘motion correction.’ The aim is to remove movement artifact in fMRI and PET time-series. SPM does this by realigning a time-series of images acquired from the same subject using a least squares approach and a 6 parameter (rigid body) spatial transformation. 

a)      Select Realign from the SPM button window. 

b)      Choose “New” "Realign and Unwarp ”

c)      On the left part of the graphics window is where you will specify a tree of data processing options, including the data. 

i)        If there is a + on the left of a branch, you can double click the branch to see the sub-branches

ii)       If there is an X on the right of a branch, then you MUST specify that option for the tree (or else you will not be allowed to RUN).

iii)     If you accidentally add an option you do not want, there is an options panel in the upper right potion of the graphics window where you can remove the extra option.

d)      Single click “Data<-X” 

i)        Single click “New”Session”

(1)   Single click “Session <-X”.

(2)   Do not specify the 'phase map'--only click on 'Images'

(3)    Then single click “Specify Files.” A Session window pops up. In the Session window, under “Drive,” click on the drive that your .img files are stored. Single click the folders (or path) to where your data is stored. Once in the folder (where the data is), highlight ALL of your .img files (using the shift key) and all your files will drop down to the bottom of the Session Window. Hit DONE. 

(4)   In the bottom right part of the Graphics window, there is a “Run” button. Click the “Run” button to start Realignment. 

(a)    New images will be saved into the same directory where the functional images are located. These new images will be prefixed with the letter ‘r’. SPM will then plot the estimated time series of translations and rotations. SPM will also create a mean image (eg. meanfM93847_002.img) which will be used in the next step: coregistration. 

 

3)      Coregistration: A Process to overlay structural and functional images in a way that maximizes the mutual information. The mean functional image is coregistered to a high resolution anatomical image, and all of the other functional images are then resliced to align with the reference image.

a)      Click the “Coregister” button in the Button Window

i)        “+Coreg: Estimate & Reslice <-X” is newly created. Double click on it.

e)      Highlight "Other Images" and select all of the images with the prefix 'u' before it (the files created in the realign and unwarp step). Do not worry about the estimation options.

g)       Press the Save button and save the job as coreg.job

h)       Press Run.

i)       Press the “Check Reg” button in the Input Window and then select the Reference and Source Images specified above (eg. meanfM93847_002.img and sM00354_003.img). SPM will then produce both images in the graphics window where you can confirm their anatomical correspondence.

 

4)      Normalize: Warp to fit to standard template brain (Montreal Neurological Institute template), in order to make inter-individual comparisons and combine data from several subjects. Not necessary if only processing data from a single subject.

a)      Click the Normalize button in the Button Window

b)      In the Graphics Window, select “New”Normalise: Estimate & Write”

i)        Double click “Normalise: Estimate & Write <-X”

ii)       Click “Data <-X” – highlight “New”Subject”

(1)   Double click “+Subject<-X”

(a)    Highlight “Source Image <-X” – Highlight “Specify Files” and select the structural image (same structural image as previous section.).

(b)   Highlight “Images to Write <-X” – Highlight “Specify Files” and select the mean functional image from the previous section and the realigned/unwarped images (with the 'u' in front).

(2)   Double click “Estimation Options <-X”

(a)    Highlight “Template Image <-X” and then Highlight “Specify Files”

(i)      Select the “T1.nii,1” file under the subfolder “templates” in the folder, SPM5b and then hit “done.”

iii)     Hit Run. 

 

5)      Smooth: This module convolves images with an isotropic Gaussian Kernel (3-D normal curve). 

a)      Press the “Smooth” button in the Button Window and then double click “+Smooth<-X)

i)        Highlight “Images to Smooth <-X” – click “Specify Files”

(1)   Select images to be smoothed (all of the spatially normalized files created in the last section (e.g. wrfTJ031908H1.img). 

(2)   Optional: Highlight “FWHM” – Highlight “Specify Text”

(a)    You can change how much you want to smooth your data. We used the default-- 8mm. (Friston et al., 2000. discusses how much you can smooth: 4 – 12mm in each direction; typical to use 6-8mm in fMRI).

(3)   Save the job as smooth.mat and Press “Run.” Final smoothed images will have a ‘s’ added to the prefix.