The lecture will outline the current situation of SHFJ group related to the automatic management of cortical folding. The whole system designed by our groupis a symbolic alternative to the usual deformable atlas principle. This alternative consists of using a higher level of representation of the data to overcome some of the difficulties induced by the complexity and the striking variability of the cortical folding. The approach is made up of two stages. First, a sequence of automatic algorithms extract from any T1-weighted MR image an abstract structural representation of the cortex folding patterns. The structural representation is a graph whose nodes represent elementary folds and whose links represent relationships between these folds. This graph is supposed to include all the information required by the sulcus recognition. Each graph node is given a sulcus name by an automatic labelling method. This pattern recognition method may be interpreted as a graph matching approach. The matching is driven by the minimization of a global function made up of local potentials. Each potential is a measure of the likelihood of the labelling of a restricted area. This potential is given by a multi-layer perceptron trained on a learning database. A base of 26 brains manually labelled by a neuroanatomist is used to validate our approach. The whole system developed to date for the right hemisphere is made up of a congregation of 265 neural networks. This system is used for epilepsy surgery planning in Marseille.