Electronic Thesis and Dissertation Repository

Thesis Format



Master of Science




Johnsrude, Ingrid S.


Treating drug-resistant epilepsy with surgery requires the localization of the epileptic focus. We explored the potential for movie-driven functional magnetic resonance imaging (fMRI) to act as a sensitive, non-invasive, and cost-effective tool to identify functionally disturbed networks. We assessed neural synchronization (inter-subject correlation; ISC) between presurgical epilepsy patients (n = 18) and healthy controls (n = 24) as they watched a suspenseful movie clip in the scanner. To optimize denoising, we compared ISC values with and without an automated Independent Components Analysis-based denoising step (ICA-AROMA). We found that denoising with ICA-AROMA elicited augmented correlation values, supporting its use for denoising naturalistic fMRI data. We identified abnormal overall ISC profiles in five of 18 patients and also observed region- and patient-specific ISC abnormalities. Naturalistic fMRI should be further explored for its utility as a sensitive and reliable complement to standard epilepsy surgical planning tools, potentially leading to improved treatment and outcomes.

Summary for Lay Audience

Epilepsy is one of the most common chronic neurological disorders, affecting 1% of Canadians. Epilepsy is typically treated with medication, although nearly 1 in 3 people with epilepsy are resistant to medication. For these individuals, surgical intervention involving the removal of the parts of the brain thought to cause seizures may be a viable treatment option. Identifying the origins of epileptic activity is complicated by recent evidence that suggests seizures originate from within networks of interconnected brain regions, rather than from single discrete brain regions. It is possible that naturalistic functional magnetic resonance imaging (fMRI), which involves exposing subjects to complex audiovisual stimuli like movies while they undergo brain imaging, may be useful for identifying impaired networks in epilepsy. Research is needed to establish the clinical utility of naturalistic fMRI in presurgical evaluations of epilepsy.

In this investigation, we recruited 18 individuals with drug-resistant epilepsy being evaluated for possible surgery, and 24 healthy controls. Participants watched an 8-minute suspenseful movie clip while we recorded their brain activity with fMRI. To examine differences between patients and controls, we used a measure called inter-subject correlation (ISC), which provides a correlation value reflecting the extent of synchronization of neural activity at each brain region across people watching the same movie. By using this technique, we found abnormal overall ISC profiles in five out of 18 patients. Further, we identified regions that were particularly sensitive to abnormalities, and observed significant abnormalities in seven of 18 patients. We hoped that special software that accounts for head movements in the scanner would give us better results. We found that denoising with one such software known as ICA-AROMA resulted in larger correlation values, supporting its use with naturalistic fMRI data, particularly when the goal is to identify neural abnormalities in clinical samples. Overall, we anticipate that naturalistic fMRI paradigms will continue to be explored for their possible utility as sensitive and reliable complements to the surgical planning tools currently used in the presurgical evaluation of epilepsy, potentially leading to better treatment and improved outcomes.