Cytoarchitectonic investigations are based on the analysis of structural inhomogeneities in the neuronal composition of nervous tissue. Boundaries of brain regions are established at locations where local structural properties, such as density, shape, orientation or arrangement of the nerve cells, change. A quantitative approach requires a complete scanning of histological sections and the measurement of at least one of these properties using an automatic device. In contrast to recently developed stereological methods which result in unbiased estimates of parameters and which are based on efficient sampling procedures, measurements with automatic devices are biased due to section thickness and problems in the segmentation of nerve cell bodies. Automatic measurements are necessary, however, if a complete scanning of histological sections is required in order to localize areal or laminar boundaries. In this approach, the grey level index (GLI) is measured with a TV‐based image analysing system from routine histological sections. Using Nissl‐staining, this parameter is a biased estimate of the local volume density of Nissl‐positive structures (cell bodies). The histological section is digitized into a GLI image by a scanning procedure. The GLI image is processed by image enhancement procedures in order to visualize the laminar pattern. Areal boundaries are localized at positions where this laminar pattern changes. GLI statistics of single brain regions can easily be evaluated by delineating these regions with a cursor in the GLI images of the sections. Information from a series of sections is compiled by specific application programs. 1990 Blackwell Science Ltd
