Current Research: Epilepsy
Epilepsy is a chronic neurological disease that affects nearly 60 million people worldwide and is defined by recurrent and unprovoked seizures. If the patient’s seizures are not controlled with anti-epileptic medications, they often lose their independence, and can experience profound behavioral, psychological, cognitive, social, and financial burdens. Currently, diagnosis and treatment of epilepsy involves visual inspection, channel by channel, of electroencephalographic (EEG) signals recorded noninvasively from the scalp or invasively through intracranial monitoring. These methods of EEG assessment are extremely time-consuming, subjective, and often disregard the network phenomena that drive seizure genesis. The Neural Signal Processing and Modeling Lab focuses on the development of novel computational tools that investigate epilepsy with quantitative frameworks to characterize the disease more objectively and accurately. Incorporation of such computational tools into clinical practice will advance current clinical decision support systems, and has the potential to significantly improve treatment outcomes.
Project: Cortico-cortical evoked potentials to identify epileptogenic (“resonant”) regions in the brain
We aim to identify epileptogenic brain regions in patients with focal epilepsy to improve surgical outcomes. Cortico-cortical evoked potentials are elicited with single-pulse electrical stimulation. We can build directed connectivity networks by analyzing the responses to stimulation in different brain regions and use properties of those networks to identify regions with pathologically high cortical excitability. We build dynamical network models from the intracranial EEG data recorded during stimulation, and with those models predict frequencies that might be “resonant” for that area. With these resonant frequencies, we aim to elicit the patient’s native seizure activity, which may provide an extra test of specificity before resection of the given brain area. This project is funded by the American Epilepsy Society (2023-2024).
Keller et al 2014; Hays et al 2021
Project: Virtual stimulation of EEG and MEG networks
Electrical stimulation provides insights into brain regions with high cortical excitability. Preliminary results show that virtual stimulation in a dynamical network model built from interictal EEG data may also identify hyperexcitable regions — without the need for electrical stimulation at all. Further, we will test whether or not we can obtain complementary clinical information by virtually stimulating a dynamical network model built from MEG data, a completely non-invasive measure of brain activity. This could aid in the localization of pathological activity so that depth electrodes could be placed in the most crucial regions of the brain for invasive monitoring and possibly eventually removing the need for intracranial investigations at all. This project is funded by CURE Epilepsy Foundation (2023-2024).
Project: MNE-HFO
High-frequency oscillations have been shown to be important neural features that may highlight epileptogenic brain regions. We are developing a Python-based platform to first try multiple HFO detection algorithms and then compare across the set of available algorithms in several large, open-source datasets.
Project: Quantitative EEG biomarkers of SYNGAP1
SYNGAP1 is a genetic disorder that causes severe intellectual disability, sleep problems, hyperactive disorders, and epilepsy. We have obtained EEG data from SYNGAP1 and neurotypical children while they listened to sets of tones. Quantitative features of the EEG when the brain is processing auditory stimuli may distinguish children with SYNGAP1 from healthy children. These features could be used as biomarkers of treatment response.
