fMRI data preprocessing¶
Normalisation¶
Normalisation refers to bringing your data into a standard template space to allow group-level statistics.
Why normalise my data?
Normalisation is an essential step in neuroimaging analysis that allows group-level analyses. When neuroimaging data is normalised it means it has been transformed to fit into a standard template space. While many different templates exist, the most commonly used one is MNI152.
For a thorough overview of issues related to normalisation, see the SPM book:
And other readings:
Ashburner, J. and Firston, K.J. (2005) Unified segmentation.
-
From the SPM menu panel, select
Normalise (Write)
. You will see a pop-up window appear looking like this: -
Select
Data
New: Subject
. - Select
Deformation field
. - In the pop-up window, use the left-hand panel to navigate to
sub-01/anat/
. - From the right-hand panel, select the deformation field generated during coregistration -
y_sub-01_T1w.nii
and pressDone
. - Select
Images to write
. - In the pop-up window, navigate to
sub-01/func/
. - From the right-hand panel, use the box underneath the
Filter
button to show a 4D file by typing inNaN
and pressing Enter. You can do this in combination with filtering for files starting withar
by typing in^ar.*
in theFilter
box and pressing Enter. Select the realigned and slice time corrected data -arsub-01_task-auditory.nii
. - Optional: Under
Writing options
, changeVoxel sizes
form[2 2 2]
to[3 3 3]
. This will write images at a resolution closer to that at which they were acquired. - Save this batch for future reference -
File
Save batch
and name it, e.g.normalisation_batch.mat.
- Run your batch by pressing .
SPM will now write spatially normalised files to the functional data directory, i.e. sub-01/func/
. These files have the prefix w
.