Skip to main content

Mapping Epileptic Networks with Virtual Cortical Resection

Brian Litt, MD, Director of the Penn Epilepsy Center and Penn’s Center for Neuroengineering
Brian Litt, MD, Director of the Penn Epilepsy Center and Penn’s Center for Neuroengineering
Brian Litt, MD, Director of the Penn Epilepsy Center and Penn’s Center for Neuroengineering

While new surgical techniques and implantable devices are increasingly available for up to 1 million people in the United States with drug-resistant epilepsy, rates of seizure freedom after surgery haven’t significantly improved.

It’s not for lack of tools, says Brian Litt, MD, Director of the Penn Epilepsy Center and Penn’s Center for Neuroengineering and Therapeutics. The missing piece is a standardized blueprint to guide their use.

“Today, a patient who goes to five epilepsy centers might be offered five different implants and five different approaches to surgery,” Litt explains. “That’s because surgical mapping is done manually by marking seizures on intracranial EEG (IEEG) by eye, and then trying to determine seizure onset time, location, and the regions of origin.”

Enter virtual cortical resection (VCR): a promising method for quantitively mapping epileptic networks and simulating interventions to better predict patient outcomes. Dr. Litt is developing this technology together with Danielle Bassett, PhD, Associate Professor of Bioengineering and the recipient of a 2014 MacArthur fellowship during her first year at Penn.

Advancing the State of the Art

Over the last four years, Litt’s team—a group of neurologists, neuro- and bioengineers, radiologists, and others—has been studying VCR to develop novel algorithms that may soon replace today’s imprecise, manual method for guiding epilepsy surgery.

This research, published in the October 2019 issue of Brain, was R01-funded and received additional grant funding from the National Institutes of HealthJonathan Rothberg, and the Mirowsky Family Foundation. The brain network model was validated on a retrospective cohort of 28 patients with drug-resistant epilepsy who received intracranial electrodes prior to surgical resection. With their outcomes in hand, Litt and his team compared their model to the patients’ outcomes.

“We’ve had quite a bit of success so far,” Litt says. “Using a simulation to ‘treat’ different areas in a patient’s brain, the model accurately pinpointed places that were good candidates for ablation, resection, or modulation with an implantable device. It’s really a game-changer.” 

Litt identifies additional benefits of virtual cortical resection. First, VCR could end the practice of provoking seizures in patients to map epileptic networks. Furthermore, it should reduce sampling bias and error, which are looming challenges to IEEG and epilepsy interventions.

“Knowing whether implants have missed vital seizure generators is key,” Litt says. “Together with Dr. Bassett’s team, we’re developing methods to assess the robustness of network models to under-sampling and to predict missing information.”

Validation by Crowd-Sourcing

With what appears to be a viable process, the team’s next goal is to generalize and validate the methods and port them to clinical trials as quickly as possible.

This will require solving the major practical obstacles to validating and translating the method into clinical care. “Two major goals are to eliminate sampling and bias errors,” Litt explains, “and to validate the methods on a larger number of patients across centers. We also want to adapt our methods to Stereotaxic EEG.”

To clear these hurdles, Litt and his team have turned to epilepsy centers across the country.

“We’ve assembled a group of 15 major epilepsy centers to collaborate with us on validating the method,” he explains. “This larger group will aggregate and share data, algorithms and code in order to better position this method for a prospective clinical trial and patient care. The response from leading epileptologists has been enormous.”

Penn Medicine, Philadelphia, PA 800-789-7366 © , The Trustees of the University of Pennsylvania

Powered by BroadcastMedBROADCASTMED