A Hybrid Approach to Quantify Lamination of the Cerebral Cortex

摘要

Alterations of cytoarchitectonic (CA) patterns in digital images of histological sections of the human cerebral cortex (CX) might indicate changes of brain functions. These lamination patterns can be extracted by calculating orthogonal testlines through CX by a stepwise scanning. 3D CA data of whole human brains at a histological resolution are not available and 2D sections deliver partial information only. The problem is to find an optimal method of scanning producing a minimum of distortions and noise by the transformation of the curvilinear CX to a rectangular presentation of CA layering. So far, the cortical surfaces are modeled as electrically charged regions to use the resulting field lines as testlines. However, local information of cell distributions was not considered. A hybrid approach constructs significantly better testlines in CX images with mixtures of columnar rich (local orientation rich) and orientation poor parts of strongly curved regions. The new hybrid approach is based on combining streamlines of the electric field with the structure tensor and constrained anisotropic diffusion. In addition, the introducing of projective transformations yields a significant improvement of cortical fingerprints. The statistical evaluation of the new method turns out to be robust with respect to artifacts and low local orientation information. The new technique can be generalized and applied to many different types of CX with local orientation information.