Datensatz

Zusammenfassung

Improvements in the spatial resolution of structural and functional MRI are beginning to enable analysis of intracortical structures such as heavily myelinated layers in 3D, a prerequisite for in-vivo parcellation of individual human brains. This parcellation can only be performed precisely if the profiles used in cortical analysis are anatomically meaningful. Profiles are often constructed as traverses that are perpendicular to computed laminae. In this case they are fully determined by these laminae. The aim of this study is to evaluate models for cortical laminae used so far and to establish a new model. Methods to model the laminae used so far include constructing laminae that keep a constant distance to the cortical boundaries, so-called equidistant laminae. Another way is to compute equipotentials between the cortical boundary surfaces with the Laplace equation. The Laplace profiles resulting from the gradients to the equipotentials were often-used because of their nice mathematical properties. However, the equipotentials these Laplacian profiles are constructed from and the equidistant laminae do not follow the anatomical layers observed using high resolution MRI of cadaver brain. To remedy this problem, we introduce a novel equi-volume model that derives from work by Bok (1929). He argued that cortical segments preserve their volume, while layer thickness changes to compensate cortical folding. We incorporate this preservation of volume in our new equi-volume model to generate a three-dimensional well-adapted undistorted coordinate system of the cortex. When defined by this well-adapted coordinate system, cortical depth is anatomically meaningful. We compare isocontours from these cortical depth values to locations of myelinated bands on high-resolution ex-vivo and in-vivo three-dimensional MR images. A similar comparison was performed with equipotentials computed with the Laplace equation and with equidistant isocontours. A quantitative evaluation of the equi-volume model using measured image intensities confirms that it provides a much better fit to observed cortical layering.