To map out seizure-inducing zones in the brain with greater accuracy, scientists have designed a virtualized brain model. This model is based on imaging and electrical activity data from patients who are resistant to drugs for epilepsy. The digital workflow, created by Huaifang Wang and colleagues, could help improve patient outcomes by informing more precise surgical procedures and interventions for epilepsy.
Epilepsy is one of the most common neurological disorders, and cases that don’t respond to drugs must instead be treated with surgery. However, surgeons must first identify the brain regions that generate epileptic signals (epileptogenic zones) with pinpoint accuracy. Recently, research has suggested that epileptogenic zones consist of hierarchical networks, rather than a single focus in the brain.
Virtualized brain is supported by machine learning
To pinpoint which regions generate epileptic signals, Huaifang Wang and colleagues created a virtualized brain model called Virtual Epileptic Patient (VEP) workflow. The researchers constructed their virtualized brains with synthetic and MRI data and electroencephalography recordings from individual patients. This data is used to build network nodes equipped with their own neural models to simulate seizure activity. It also combines personalized brain models with machine learning techniques to estimate epileptogenic zone networks and to aid in clinical decision-making.
The workflow allowed the team to simulate various “virtual surgery” treatment strategies for several patients. It showed good precision when retrospectively compared to the clinical hypothesis in a group of 53 patients. “Both scientists and clinicians are currently searching for equally efficient non-invasive approaches for diagnosis and treatment in the future for patients with epilepsy,” Wang et al. say. “Toward this purpose, the whole-brain network modeling could play an important role in diagnostic and therapeutic solutions.”
This complete paper is available here: www.science.org/doi/10.1126/scitranslmed.abp8982
