Establishing the Language Network With Independent Component Analysis of Resting State MRI

Patients with brain tumors including malignant gliomas located in eloquent areas have long been considered inoperable because of median survival and predictable postoperative sequelae. The improved prognosis with chemoradiotherapy and surgical resection as complete as possible questioned this attitude. The goal is to operate those patients without additional disabilities. The cortical electrostimulation intraoperative mapping in awake surgery can meet the eloquent areas. Identifying with fMRI these same areas to better specify in advance the way initially and shortened the time to awake exploratory surgery which improves patient comfort. However anxiety induced by the review in a pre-operative setting, a debutante aphasia often make problematic the realization of a formal language test. The analysis in the low frequency of spontaneous oscillations of large neuronal networks can help isolate network of language at rest without particular language test at the group analysis. This methodology, however, needs to be validated at the individual level before it can be applied to pathology.

Proceed to an analysis of sensitivity and specificity results of the fMRI language network obtained in two ways: by providing a conventional language test and independent component analysis. These two methods will be compared to the reference method what the cortical intraoperative mapping.