Functional magnetic resonance imaging (fMRI) has become a popular technique for studies of human brain activity. Typically, fMRI is performed with >3-mm sampling, so that the imaging data can be regarded as two-dimensional samples that average through the 1.5-4-mm thickness of cerebral cortex. The increasing use of higher spatial resolutions, <1.5-mm sampling, complicates the analysis of fMRI, as one must now consider activity variations within the depth of the brain tissue. We present a set of surface-based methods to exploit the use of high-resolution fMRI for depth analysis. These methods utilize white-matter segmentations coupled with deformable-surface algorithms to create a smooth surface representation at the gray-white interface and pial membrane. These surfaces provide vertex positions and normals for depth calculations, enabling averaging schemes that can increase contrast-to-noise ratio, as well as permitting the direct analysis of depth profiles of functional activity in the human brain.
Bibliographical noteFunding Information:
We thank the VISTA lab at Stanford University, particularly Bob Dougherty and Brian Wandell, for their assistance in developing some of the visualization tools and fMRI analysis methods described here. We are also grateful to Gary Glover for providing the MRI spiral-acquisition pulse sequence. Research for CB was supported in part by NIH Grants R01EB00487, R01GM074258, and R01GM07308.
- Deformable surface
- Surface models