Résumé
For a long time, brain functioning was conceived in a localizationist and static workframe. This view was source of limitation for cerebral surgery, because according to this dogma, resection of a lesion involving an area a priori considered as "critical" would generate irrevocable neurological deficits. Recents developments enabled a switch towards dynamic organization of brain processing, based upon complex and interconnected networks able to compensate themselves following cerebral injury. Indeed, it is possible to perform individual functional mapping, thanks to both functional neuroimaging as well as brain surgery under local anesthesia: intraoperative awake mapping allows the indentification and preservation of cortical and subcortical eloquent structures. Advances in cerebral connectomics also opened the door to neuroplasticity, which dramatically changed the neurosurgical philosophy. Therefore, one could now consider to remove brain tumors in regions classically considered as "unoperable" with a minimal risk of neurological worsening, while increasing the extent of resection and thus overall survival: the principle of "functional neurooncology" is born. These concepts begin to be applied to epilepsy surgery, with the main goal of optimizing quality of life. A better understanding of neural networks also resulted in neuromodulation, through deep brain stimulation, which was demonstrated as very efficient in movement disorders (as Parkinson's disease), and growingly used in chronic pain or psychiatric diseases (e.g. severe depression, obsessional compulsive disorders,...). In summary, neurosurgery shifts toward a connectionist view, based on an improved knowledge of dynamic cerebral circuits, and opening new therapeutic avenues in the field of "neural networks surgery".