Summary:
UCLA researchers in the Departments of Neurology and Electrical and Computer Engineering have developed a data-driven, deep learning-powered software platform to localize epileptogenic tissues in the brain and cognitively map nearby regions.
Background:
According to the World Health Organization, 50 million people around the world suffer from epilepsy with 1.2 million in the US alone. While medications can help some, about 30% of these patients have a drug-resistant form of the disease. Focal seizures, also known as partial seizures, can be treated via electrode implantation in the brain or tissue removal at the regions where the seizures originate. Both treatment options carry significant risk in the opportunity for infection or inadvertent removal of tissue important for normal cognitive function. There is a clear need for treatment options that facilitate high precision in localizing epileptogenic tissue in the brain as well as determination of cognitive function in the relevant regions.
Innovation:
UCLA researchers have developed a data-driven, deep learning-powered software platform trained on scans from multiple information sources including fMRI and MRI scans, PET scans, and EEG data. Data from these modalities can be complex to interpret especially when extracting relevant data in a noisy environment, so training artificial intelligence to reduce the time required benefits both patients and physicians. Patient response to treatments will also be incorporated in training of the models, enabling more precise and effective recommendations. This software also includes a user-friendly interface which simplifies use to enhance patient outcomes. Ultimately, this innovation could allow users to generate a heatmap to visualize regions of the brain most likely to correlate with seizures and further train a deep learning model to better detect these regions of higher risk.
Potential Applications:
• Localization of epileptogenic tissues in the brain
• Cognitive function mapping of brain tissues
• Training of neurologists and radiologists
Advantages:
• Better patient outcomes in tissue resection
• AI driven platform
• Less invasive identification of epilepsy markers in the brain
Development-To-Date:
The system has been described in full (March 2023).
Related Papers:
Application of an EEG-based deep learning model to discriminate children with epileptic spasms from normal controls. July 01, 2023. Mingjian Lu, Yipeng Zhang, Atsuro Diada, Shingo Oana, Rajsekar R. Rajaraman, Hiroki Nariai, Vwani Roychowdhury, Shaun A. Hussain. doi: https://doi.org/10.1101/2023.06.30.23292096
Reference:
UCLA Case No. 2023-214
Lead Inventor:
Vwani Roychowdhury