Brodman divided the cortical surface into 47 areas on the basis of gross morphology (sulci) and cytoarchitechtonic studies. In broad terms, this division relates gross anatomy (such as can be seen by conventional MRI) to the localization of function.

The cortex is not uniform, and its thickness varies by region. Cortical thickness tends to increase as one goes from primary sensory cortex to sensory association cortex to heteromodal cortex to motor association cortex to motor cortex.

Studies at the microscopic level indicate that the number of neurons and cortical columns in various brain regions is more related to the surface area of that brain region than to its volume.

The cortex is a relatively thin laminar structure, so that the volume of a given cortical region is roughly proportional to its surface area times its thickness. Thus, given any two of three values for volume, surface area, and thickness, it is possible to deduce the third.

Many investigators studying neuropathology and development use MRI techniques to measure the volume of various cortical regions. The purpose of this research effort is to extend current measurements of cortical regions from volume measurements to measurements involving the more biologically meaningful surface area. The figure shows data from a model in which cortical thickness is deduced from a two-dimensional histogram which is a generalization of the conventional one-dimensional image histogram. In this histogram, the second dimension relates to the normal distance from the center of the voxel to the gray-matter/white-matter manifold. Under assumptions dictated primarily by geometric concerns, the contents of this two-dimensional histogram can be predicted by a simple model, and the parameters of this model (which include local thickness) can be fit by simple Bayesian techniques. Given this value for thickness, and other geometric parameters fit by the model, it is also possible to predict both the surface area and volume of the region. One notable feature of the model is that it does not explicitly require the location of the gray matter/cerebrospinal interface. In practice, this is an advantage because this interface is currently impossible to visualize throughout the entire brain given existing MRI technology.