Abstracts

Rationale: Functional MRI (fMRI) is widely used to map functional areas in the brain, but its applications to mapping brain areas involved in epileptogenic pathology have been less well-studied. Recent work has applied resting fMRI data to parcellate the brain into functional networks and their modular components (i.e., cortical areas) based upon intrinsic functional connectivity (iFC) patterns detected during rest. If these parcellations correspond to borders between functional and epileptogenic brain regions, they have the potential to non-invasively guide the resolution of seizure onset areas from functional brain areas, and inform surgical planning for focal epilepsy. We used gradients derived from intrinsic functional connectivity (iFC) similarity (Wig et al., 2014) to distinguish functionally distinct cortical areas within brains of individual subjects (Xu et al., 2016) of patients undergoing presurgical monitoring for drug-refractory epilepsy. We then tested these gradient-based parcellations of brain areas against clinical identifications of functional (language, motor) and seizure onset zones obtained from invasive monitoring, to determine correspondence of gradient-based parcellations to functional and epileptogenic brain areas. Methods: For 3 patients with focal neocortical epilepsy, gradients were derived from intrinsic functional connectivity (iFC) similarity and used to define edges for a functional parcellation of brain areas within each subject (Xu et al., 2016). These parcellations were compared directly to functional and epileptogenic brain areas identified in the course of invasive mapping. Areas of function were defined using data from clinical stimulation mapping (CSM) of implanted patient electrodes. For this analysis, assessment of brain function was restricted to language (including speech arrest, picture naming, auditory naming) and motor phenomena. Areas involved in seizure onset were identified using epileptologist review of ictal intracranial electroencephalography (iEEG).  Results: Gradient-based resting fMRI parcellations corresponded to either a functional or an epileptogenic zone of the brain identified in the course of intracranial monitoring for each of the 3 patients studied. In one patient, gradient-based parcellations correspond to edges of the seizure onset zone as defined by invasive monitoring (Figure 1). We find that gradient-based resting fMRI parcellations can identify regions of function and epileptogenic pathology in the brain in this limited sample. Conclusions: Gradient-based resting fMRI parcellation provides a promising method to identify seizure onset zones as well as functional zones in the brain. Future investigation will be required to optimize correspondence of gradient-based resting fMRI parcellations with data acquired from invasive monitoring, particularly iEEG. We suggest this may be accomplished using machine learning methods. Funding: NIH 1R01-MH111439-01.