ABSTRACT

Since the discovery in the early 1990s, functional MRI (fMRI) has been increasingly utilized as a technique to investigate human brain function. The most direct clinical application in which fMRI is already playing an important role is presurgical mapping for patients with brain tumors near functional cortical areas. The goal of preprocessing is to reduce artifactual signal variance in the voxel time series that is not associated with the subject’s functional task and to improve the detectability of neurally induced changes. Preprocessing steps generally encompass image reconstruction, slice-timing correction, motion correction, spatial smoothing, and spatial normalization. The “slice timing” problem refers to a continuous ascending/descending gradient echo-planar imaging sequence in which, for example, the top slice is acquired at a time equal to the repetition time later than the bottom slice. Head motion during the fMRI experiment poses the biggest practical problem, especially in clinical fMRI. Three-dimensional volume registration is generally useful for intra- and intersession alignment.