Awareness of the critical
importance of fMRI pre-processing is increasing for both task-based and
especially resting-state fMRI research. Most resting-state studies address
questions of functional connectivity, i.e. target the correlation of brain
activity in one area with activity in a different brain area. This means that
regressors used in first-level linear models of resting-state fMRI come from
the brain itself, rather than from externally generated task designs that are
unaffected by acquisition artifacts or pre-processing steps in task-based fMRI.
In contrast to task-based fMRI, independent and dependent variables are thus
both affected by artifacts and pre-processing steps, and there is a greater
chance of artificially induced functional connectivity than task-based
activation. It follows further that those common pre-processing pipelines which
have gained acceptance in task-based fMRI practices should not necessarily be
carried over to resting-state studies of functional connectivity. After attending
our proposed educational course the audience should have gained a thorough
understanding (1) of the kinds of artifacts are affecting the hemodynamic
signal recorded in fMRI scanners and (2) of the state-of-the-art tools to
counteract these artifacts. Beyond these initial learning objectives, course
attendees should have gained awareness of the problem of pipeline dependence
and the ability to follow, and possibly engage in, methodological research that
aims at pipeline optimization using real-world as well as simulated data.
Feedback from the previous 2 years when this course was held was incorporated:
(1) practical aspects of familiarization with the three major software packages
(SPM,AFNI, FSL) were incorporated; (2) an emphasis on pre-processing as an important,
as yet “unfinished”, area of methodological research of vital importance for
the integrity of neuroscience at large.