A recently published paper from a team of researchers and students from multiple departments at the Cullen College of Engineering and the McGovern Medical School at UTHealth proposes a way to provide real-time, continuous tracking for the 50 million people living with potential seizures from epilepsy world-wide.
While a drug regimen can control many symptoms of epilepsy, according to the paper's authors more than 90 percent of people with the disorder will still experience seizures. With real-time monitoring, a closed-loop system can be developed – making sure a person is only given treatment at the most effective times, to prevent or to cut off seizures.
The research was truly a collaborative effort between multiple departments at the college and outside of it, as the paper has nine listed authors, eight of which are at UH.
The senior author of the paper is Rose T. Faghih, Ph.D., Assistant Professor of Electrical and Computer Engineering, and a member of the BRAIN Center. The first author for the paper is Alexander Steele, a doctoral student of Jose Luis Contreras-Vidal, Ph.D., Hugh Roy and Lillie Cranz Cullen Distinguished Professor and Director of the NSF IUCRC BRAIN Center. Steele and fellow student author Sankalp Parekh wrote the paper as part of a course, “State-Space Estimation with Physiological Applications,” taught by Faghih.
Steele and Parekh built on the previous work of course students Mohammad Badri Ahmadi and Alex Craik. Hamid Fekri Azgomi, a doctoral student of Rose T. Faghih, was the course TA. Ahmadi is a doctoral student of Joseph T. Francis, Ph.D., professor of Biomedical Engineering.
From outside UH, Sandipan Pati, MD – an associate professor specializing in epilepsy at the McGovern Medical School at UTHealth – also contributed to the research.
The paper, “A Mixed Filtering Approach for Real-Time Seizure State Tracking Using Multi-Channel Electroencephalography Data,” was published in September in Volume 29 of IEEE Transactions on Neural Systems and Rehabilitation Engineering.
According to a summary of the research provided by Steele, “The proposed algorithm uses electroencephalography (EEG) data, which is electrical activity from the brain recorded non-invasively from the scalp. We use EEG measurements originated from different channels as the observation to derive multiple estimations. Next, we utilize a mixed filter approach and employ both binary and continuous features to estimate the seizure state. This estimate not only predicts if there is a seizure occurring, but importantly it predicts the probability of the seizure occurring. The continuous outcome would further help the medical doctors to track how well treatments or medications work at preventing/reducing seizure activity.
Using a Kalman filter we can then employ multiple estimations and weigh them based on how well they capture seizure activity. Consequently, the proposed method is effective in detecting the general seizure that could be triggered in multiple parts of the brain. In addition, as the EEG signal is vulnerable to be contaminated with movement artifacts or other sources of noise, the idea of combining the estimates from different channels would further enhance the robustness of the proposed approach. The proposed architecture is also computationally efficient, which is important if this will be applied to portable devices. We believe accurately estimating seizure states may guide treatment and improve quality of life for people who have epilepsy.”
Steele noted that Faghih provided essential framework for him and Parekh in the course, when it came to research.
“She was careful to give well thought out examples and practice problems,” he said. “The material covered introduced me to a wide variety of useful techniques. More importantly, I was taught how I could apply those techniques to problems not explicitly covered, which has already proven to be an invaluable lesson.”
Steele called the integration of the research project and classwork “seamlessly” done.
“The mentorship I received while working on this project was very hands on and detailed. I learned a lot from both the faculty involved and the other students whom I got the chance to work with,” he said. “In the end, this project was published, which is very rare considering this was all done for a class. I learned a remarkable amount about how to best write and format a paper. If it weren’t for the mentorship, I received none of it would have been possible and the experience made me a better researcher. I firmly believe this class has provided the most growth for me as a student since I started the Ph.D. program, and I highly recommend it to other students.”