Tasklists in aa (version 4)
- aap_tasklist_typical_fmri.xml – standard processing pipeline, described in more detail below
- aap_tasklist_noslicetiming.xml – standard pipeline, no slice timing
- aap_tasklist_noslicetiming_cbu32channel.xml – no slice timing, and offset applied to structural prior to normalization
- aap_tasklist_structuralsonly.xml – converts structural and normalizes it
- aap_tasklist_dartelvbm.xml – uses Dartel for VBM pre-processing
- aap_tasklist_dartelvbm8.xml – as dartelvbm but for SPM 8
The tasklist is specified in the recipe. In aa version 4, you can only specify one parameter in the aarecipe command, which is taken as the tasklist.
The processing pipeline set up by aap_tasklist_typical_fmri.xml is shown in the graphic below
In aa version 3, or less commonly for special requirements in version 4, you may also want to change the default parameter file. You do this by providing two arguments to aarecipe
The structural and EPI (echo-planar imaging) data are initially stored in DICOM format (.dcm) and split according to subject and sequence. These modules help convert the scanner’s DICOM-files to NIFTI-files (.nii or .img/.hdr) for neuroimaging analysis. A pre-specified directory and its sub-directories will be specified and searched for the appropriate DICOM-files and subsequently converted to NIFTI format using SPM functions. These NIFTI files will be appropriately named and sorted according to their type (i.e. anatomical (aamod_copystructural), functional (aamod_convertepis) or DTI). The moduleaamod_converttmaps can also be used to convert the resulting statistical maps from DICOM to NIFTI format.
aamod_copystructural / aamod_convertepis aamod_coreg_noss / aamod_realign / aamod_tsdiffana
When the head moves during an experiment some of the images will be acquired with the brain in the wrong location. The goal of motion correction (aamod_realign) is to adjust the time series of images so that the brain is in the same position in every image. The general process for spatially aligning successive image volumes in the time series, as well as the mean fMRI volume (reference volume), with the structural scan is called coregistration (aamod_coreg_noss). We hope that the brain is the same in every image of the time-series and as such a rigid-body transformation is used. Rigid-body transformations assume that the size and shape of the two objects being coregistered are identical and that one can be superimposed onto the other by a combination of three translations (moving the entire image volume along the X, Y and Z axes) and three rotations (rotating the entire image volume through the X-Y-Z and Y-Z planes). The goal of motion correction is to find the rigid-body transformation at which the smallest cost function is obtained and so these diagnostic steps (realignment, coregistration), (outputted by the module aamod_tsdiffana) should be checked carefully. (See below and aa_report for further details).
The module aamod_tsdiffana also outputs information about the movement variance broken down across slices. Most fMRI data are acquired using two-dimensional pulse sequences that acquire images one slice at a time, the use of spatial gradients limiting the influence of an excitation pulse to a single slice within the brain. A typical pulse sequence will acquire a certain number of slices (24 slices+) that cover the entire brain within a pre-specified TR (repetition time). The type of slice acquisition often varies (e.g. interleaved, ascending/descending), but each have their own sampling problems producing slice-time errors. The module aamod_slicetiming attempts to correct the fMRI data for differences in the acquisition time across different slices.
aamod_norm_noss / aamod_norm_write
In most experiments we want to address questions that go beyond a single participant and look at how the group performed as a whole (group-level analysis). However, even if brain activity is well localised within a single-subject, through coregistration for instance, there remains the problem of comparing activity across individuals (either in the same study or across studies). There is a wide variation in brain size, shape, orientation and gyral anatomy and for intersubject comparison to be feasible each persons brain must be transformed so that it is the same size and shape as all of the others with each brain region matching the corresponding one in a standard/average brain template (e.g. the MNI standard brain template) . This process is called normalisation(aamod_norm_noss), applying these normalised parameters to the aquired fMRI data series (aamod_norm_write).
Normalisation, transforming the MRI data from an individual subject to match the spatial properties of a structural template, is a difficult process and results in some reduction in spatial resolution. An additional possible module is aamod_smooth which explicitly reduces spatial resolution by smoothing the fMRI data in space using a three-dimensional Gaussian filter of several voxels in width (increasing the effective voxel size to 6 x 6 x 6 mm or greater). Although such smoothing reduces spatial resolution it can improve the validity of statistical tests and comparisons across participants.