Researchers at Johns Hopkins University have created a new tool that could significantly reduce the rate of misdiagnosed epilepsy by up to 70%. This innovative device enhances standard electroencephalogram (EEG) tests, which may look normal, converting them into trustworthy indicators of epilepsy.
By uncovering hidden signs of epilepsy in EEGs that seem typical, this groundbreaking tool aims to greatly lower the global false positive rate of approximately 30%. This improvement could spare patients from the adverse effects of unnecessary treatments, restriction on driving, and other complications related to misdiagnosis.
“Even when EEGs appear completely normal, our tool provides insights that make them valuable,” explained Sridevi V. Sarma, a biomedical engineering professor at Johns Hopkins and the primary author of the study. “We can achieve an accurate diagnosis three times faster because patients often have multiple EEGs before any abnormalities are detected, despite having epilepsy. A swift and precise diagnosis allows for quicker access to effective treatment.”
The results of this study have just been published in Annals of Neurology.
Epilepsy is marked by recurrent, unprovoked seizures caused by irregular electrical activity in the brain. The initial diagnosis usually involves scalp EEG recordings, which capture brain activity through small electrodes placed on the scalp.
Doctors utilize EEGs to identify epilepsy and assess the need for anti-seizure medications. However, interpreting EEGs can be challenging since these tests often pick up significant background noise, and seizures rarely occur during the typical 20 to 40-minute recording. These elements make diagnosing epilepsy subjective and prone to errors, even among experienced professionals, Sarma highlighted.
To increase the accuracy of EEG results, Sarma’s team studied the brain activity of patients during seizure-free intervals. Their tool, known as EpiScalp, uses algorithms based on dynamic network models to analyze brainwave patterns and reveal hidden indicators of epilepsy from a single routine EEG.
“If someone has epilepsy, why don’t they have seizures all the time? We hypothesized that certain brain regions serve as natural brakes, preventing seizures. It’s comparable to the brain’s immune response to the disorder,” Sarma clarified.
The recent study involved 198 epilepsy patients from five leading medical institutions: Johns Hopkins Hospital, Johns Hopkins Bayview Medical Center, University of Pittsburgh Medical Center, University of Maryland Medical Center, and Thomas Jefferson University Hospital. Out of these, 91 were diagnosed with epilepsy, while the rest had non-epileptic conditions that mimicked epilepsy symptoms.
Upon analyzing the initial EEGs with EpiScalp, Sarma’s team was able to identify 96% of the false positives, lowering misdiagnosis rates from 54% to 17%.
“This aspect of our tool is truly remarkable, enabling us to detect signs of epilepsy in EEGs that might appear inconclusive, thereby lessening the chance of misdiagnosis,” stated Khalil Husari, co-senior author and assistant professor of neurology at Johns Hopkins. “These patients were facing side effects from anti-seizure medications without any actual benefit, as they didn’t have epilepsy. Without an accurate diagnosis, we cannot uncover the underlying causes of their symptoms.”
Sometimes, misdiagnosis occurs due to misinterpretation of EEGs, leading doctors to overdiagnose epilepsy to prevent the possible risk of a second seizure. However, some patients have nonepileptic seizures that resemble epilepsy but can usually be addressed with non-medication therapies.
In prior studies, the team examined epileptic brain networks using intracranial EEGs and demonstrated that areas surrounding the seizure onset zone inhibit it when individuals are not having seizures. EpiScalp builds on this previous research to recognize those patterns from standard scalp EEGs.
Traditional methods for improving EEG analysis usually focus on individual signals or electrode readings. In contrast, EpiScalp looks at how different brain regions interact and affect one another through complex networks of nerve pathways, according to Patrick Myers, the principal author of the study and a doctoral student in biomedical engineering at Johns Hopkins.
“Investigating how nodes function together within the brain’s network shows patterns of independent nodes attempting to generate excessive activity while being suppressed by other regions that are not contributing to overall brain function,” Myers stated. “We check whether we can identify this pattern anywhere—do we see an area in your EEG that appears disconnected from the rest of the brain’s network? A healthy brain shouldn’t show this separation.”
The research team is now conducting a larger study to further validate its findings across three epilepsy centers and has filed a patent for the EpiScalp technology in 2023.
Other contributors to the research included Kristin Gunnarsdottir, Adam Li, Alana Tillery, Babitha Haridas, and Joon-yi Kang from Johns Hopkins; Vlad Razskazovskiy, Jorge Gonzalez-Martinez, and Anto Bagić from the University of Pittsburgh Medical Center; Dale Wyeth, Edmund Wyeth, and Michael Sperling from Thomas Jefferson University Hospital; Kareem Zaghloul and Sara Inati from the National Institute of Neurological Disorders and Stroke, NIH; Jennifer Hopp from the University of Maryland Medical Center; and Niravkumar Barot from Beth Israel Deaconess Medical Center.
The research received funding from the Louis B. Thalheimer Fund for Translational Research at Johns Hopkins Technology Ventures, as well as the National Institute of Neurological Disorders and Stroke Grant Number: R35NS132228.
