Introduction
As more and more research is done on white matter morphology, the obvious question to
ask is whether differences can be found between individuals or between groups, ie
patients vs controls. Before we begin, I’d like to summarize some important acquisition
issues.
Data Acquisition
- make sure you acquire data on each subject exactly the same. Be careful to use
exactly the same TR/TE, b-value, voxel size and number of slices.
- always measure iso-voxel data. It’s easy to understand why: when acquiring
non-isovoxel data, and if a fiber structure would be aligned parallel to the
long axis of a voxel, more contrast is added there due to the summation of
diffusivity. Ergo, more artificial signal decay in that direction!
- try to avoid interpolating your data. This has a dramatic effect on derived measures
such as FA. This is illustrated below, where the FA distribution for a native
resolution VMP versus interpolated VMP is shown for a ROI in the corpus
callosum.
- be very careful with interpretation of differences in FA or ADC. They might
not be of anatomical nature but due to the acquisition, data processing
etc.
- include a T1 weighted high-resolution anatomy in the session.
About native and Talairach space
We recommend computing the tensor and derived measures in native space (that is the
space as the data was scanned in), and to do the group analysis in common space by
converting the scalar FA/ADC maps to TAL space. This prevents introducing artefacts
in the tensor data by interpolation. Interpolation of a tensor is not trivial and
may introduce tensor deformation in the processing. For more information, see
reference [?].