Poster A52, Thursday, August 16, 10:15 am – 12:00 pm, Room 2000AB
Bringing Resting-state connectivity into the Operating Room: Comparing task and resting-state functional connectivity in presurgical language mapping
Daniel A. Di Giovanni1, D. Louis Collins1, Denise Klein1;1McGill University
Introduction: Task-based functional magnetic resonance imaging (fMRI) is typically used preoperatively to delineate eloquent cortex. Although non-invasive, this has several shortcomings as it is both labour and time intensive to find and use appropriate tasks. Resting-state fMRI (rs-fMRI) has been proposed to overcome these limitations. Previous work has shown that there is a strong relationship between the underlying network architecture of the brain during tasks and during rest. This project examines the persistence of this relationship in the language network despite the presence of brain tumors, to reinforce the notion that a fundamental intrinsic architecture of the language network drives the task-evoked activity, and hence justifies the use of rs-fMRI techniques to map the same network. This gives further credence to the potential of using rs-fMRI as an effective and efficient tool for presurgical mapping of functional cortex. Methods: Five-minute rs-fMRI scans were collected along with language task fMRI data. The language task consisted of participants naming objects and actions based on visual depictions. The data was examined in two parts: first as a group level analysis between patients and controls, and second on a subject level for 3 randomly selected patients and controls. For the group level analysis, task and resting-state data from tumour patients were compared to healthy controls. To do this, functional connectivity (FC) matrices were extracted using regions of interest (ROI) based on known areas essential for language function. A standard fMRI preprocessing pipeline was applied using the CONN toolbox, except during the denoising of the task scans, the effect of task activation was regressed out of the whole brain signal. This was done so that the resulting task-related functional connectivity matrices would be representative of the underlying functional connectivity. A weighted general linear model was used to determine significant brain activity during the task scans at the first-level of analysis. Pearson correlations with Fisher’s Z-transformed values, of each ROI to each other ROI, were used to determine functional connectivity between ROIs and create connectivity matrices for both conditions (Task and Rest) and groups (Patients and Controls). The second-level analysis correlated the rs-FC matrices to the task-FC matrices, first at a group level, and then at the individual subject level. Results: We found a high correlation between the functional connectivity matrices of the language task conditions and the resting-state in both controls and patients. Furthermore, when examined on a subject-level, the functional connectivity matrices of tumor patients correlated at similar level to that of controls. Discussion: The similarity in correlation values in controls and patients suggests that the intrinsic network architecture of the brain and its relationship to the task-evoked activity of the brain is preserved in tumor patients. This is a step forward in demonstrating the usefulness of rs-fMRI for presurgical planning even for such high-level processing such as language, as we can use this data to inform novel ways of mapping functional cortex using only rs-fMRI.
Topic Area: Language Disorders