Modelling dependencies between relations to ensure consistency of a cerebral cortex anatomy knowledge base

A symbolic model of anatomy that could be used in various contexts is a key feature. However, explicitly representing anatomy requires managing many specialisation, part-whole and topological relationships. Furthermore, we notice dependencies between some of them. These dependencies have to be taken into account in order to insure both intrinsic and incremental consistency of the model. Our approach is composed of three steps. First, we define the relationships between anatomical concepts by relationships between the space region they take up or the portion of matter they are made of. Second, we use these definitions and properties of spatial and matter relationships to infer dependencies between anatomical relationships. Eventually, we apply these dependencies to the set of independent anatomical relationships to automatically generate all the dependent relationships. This method was used to maintain an anatomical model of the frontal, temporal, parietal and occipital regions of the cerebral cortex. For 113 concepts, 221 of the 370 relationships could be automatically generated. The more the number of concepts increase, the more pertinent the method appears.