Epilepsy affects approximately 50 million people worldwide. Up to 30 per cent of patients with focal epilepsy have persistent, disabling seizures that are resistant to conventional treatment. Uncontrolled epilepsy is harmful to the brain, has devastating socio-economic consequences, and is associated with increased risk of injury and sudden death. Surgery is the most effective treatment for drug-resistant focal childhood epilepsy.

Magnetic resonance imaging (MRI) has revolutionized the evaluation and management of drug-resistant epilepsy by allowing the reliable detection of the structural lesion associated with certain focal epilepsies, thus leading to increased rates of successful resective surgery. Multiple studies have shown that the most important predictor for favorable post-surgical outcome is the complete resection of the abnormality detected on pre-operative MRI. However, despite technical improvements in MR hardware and sequences, in up to 50 percent of patients with drug-resistant focal epilepsy have a best-practice MRI that is unremarkable and thus unable to show the potential surgical target.

Our research focuses on the use of advanced MRI techniques to improve the understanding and treatment epilepsy due to focal lesions.


Connectome-informed simulations of pediatric epilepsy surgery

Surgery is the most effective treatment for drug-resistant childhood epilepsy. Yet, ~30 per cent of patients show post-surgical seizure recurrence and many suffer from cognitive side effects. So far, an outcome cannot be reliably predicted for a given patient prior to intervention.


1) Integrate preoperative MRI markers and data on the actual surgery for more effective prediction;

2) analyze brain change from pre- to postoperative time points, mapping alterations of non-resected areas; 3) simulate network reorganization using connectome models.
Hypothesis. Combining preoperative markers of patient anatomy with with models that simulate consequences of surgery will yield high accuracy in predicting clinical and cognitive outcomes.


In children that will undergo epilepsy surgery, we will obtain MRI and neurocognitive data shortly before surgery, 6 months after surgery (and 12 months after surgery. We will overlay resection extent (from postoperative MRI) with preoperative MRI features (atrophy, gliosis, myelin), testing whether anomalies outside the resection margin predict seizure relapse. In addition, fMRI pattern analysis will localize regions critical for language/memory function, testing whether their resection relates to postoperative cognitive decline. Mixed-effects models will map MRI feature change, assessing surgical consequences. Connectome-based models will be devised to predict downstream degeneration and functional reorganization after surgery in non-resected areas. Supervised learners will predict outcomes based on preoperative data and simulations.

Expected Results

Empirical studies on surgical consequences will help build effective in silico models that optimize outcome prediction, improving clinical decision-making in >10,000 children with drug-resistant seizures in Canada.
Improved detection of Focal Cortical Dysplasia.
Focal cortical dysplasia (FCD) is the most common cause of surgically-treatable, extra-temporal epilepsy. Unfortunately, the findings of FCD on MRI are often subtle and can be easily overlooked. We are using advanced MRI techniques to improve the detection of focal cortical dysplasia, allowing us to offer the life changing benefits of epilepsy surgery to more patients.


SickKids Foundation - New Investigator Grant 2016 "Connectome informed simulations of pediatric epilepsy surgery"

Research Group Members

Katerina Pezarro