Skip to main content

Event Details

  • Monday, February 25, 2019
  • 14:45 - 15:15

Epileptic Seizures: Can Deep Learning Meets Predicting Expectation?

Epilepsy is one of main neurological disorders. Less than 3% of patients only can benefit from surgery. Also, presurgical monitoring to localize the focus is a challenging step. Wearable and implantable brain-machine interfaces (BMIs) are introduced to face the localization of the seizure zone, its onset detection, and abortion of seizures before their emergence. On the other hand, on line prediction of seizures at least a half hour before its appearance is major challenges due to the complexity of brain behavior. This talk includes the description of a wearable helmet based on a fNIRS platform which is composed of an array of system-in-package based optodes proposed to measure hemoglobin variation in deep cortical levels. If a seizure is located but surgery cannot be accomplished, a multichannel mixed-signal detector and stimulator can be used to onset abort the seizures. Most important, prediction based on deep-learning algorithms is creating hope and significant expectation, that allow to prevent seizure while before its onset zone. Several research groups are conducting deep learning implementations to validate predicting algorithms using regular and intracortical EEGs to identify generators of seizure activity.