The creation of a brain atlas for image guided neurosurgery using serial histological data

Abstract

Digital atlases of the human brain can help in the specific localization of structures of surgical relevance and interest in Image Guided Neurosurgery (IGNS). This paper outlines one of the steps in the creation of a digital atlas intended for IGNS, using histological data. The acquisition of histological data can include artefacts such as tearing, shearing, stretching, shrinking, as well as inhomogeneous staining leading to structural inhomogeneities. These inconsistencies are reduced using a non-linear intensity based registration procedure where deformations are defined using a maximized correlation coefficient estimate. The intensity artefacts brought about by inhomogeneous staining are reduced by applying a slice to slice intensity inhomogeneity correction by modelling the intensity mapping between slices as a third order polynomial that is estimated with a Least Trimmed Squared fit. The lateral ventricle was then segmented and to demonstrate increased smoothing along its surface. © Springer-Verlag Berlin Heidelberg 2003.