نبذة مختصرة : Introduction: Low-frequency direct electrical stimulation is performed in drug-resistant epileptic patients, implanted with depth electrodes. It induces cortico-cortical evoked potentials (CCEP) that allow in vivo connectivity mapping of local networks. The multicentric project F-TRACT aims at gathering data of several hundred patients in a database to build a propabilistic functional tractography atlas that estimates connectivity at the cortex level.Methods: Semi-automatic processing pipelines have been developed to handle the amount of stereo-electroencephalography (SEEG) and imaging data and store them in a database. New signal processing and machine-learning methods have been developed and included in the pipelines, in order to automatically identify bad channels and correct the stimulation artifact. Group analyses have been performed using CCEP features and time-frequency maps of the stimulation responses.Results: The new methods performance has been assessed on heterogeneous data, coming from different hospital center recording and stimulating using variable parameters. The atlas was generated from a sample of 173 patients, providing a connectivity probability value for 79% of the possible connections and estimating biophysical properties of fibers for 46% of them. The methodology was applied on patients who experienced auditory symptoms that allowed the identification of different networks involved in hallucination or illusion generation. Oscillatory group analysis showed that anatomy was driving the stimulation response pattern.Discussion: A methodology for CCEP study at the cerebral cortex scale is presented in this thesis. Heterogeneous data in terms of acquisition and stimulation parameters and spatially were used and handled. An increasing number of patients’ data will allow the maximization of the statistical power of the atlas in order to study causal cortico-cortical interactions. ; Introduction : La stimulation électrique directe basse fréquence est pratiquée sur des patients épileptiques ...
No Comments.