EpiScalp could reduce misdiagnosis of epilepsy

By uncovering hidden epilepsy signatures in seemingly normal electroencephalograms (EEGs), the tool developed at Johns Hopkins University could significantly reduce false positives - seen in around 30 per cent of cases globally - and spare patients from medication side effects, driving restrictions, and other quality-of-life challenges linked to misdiagnoses.

“Even when EEGs appear completely normal, our tool provides insights that make them actionable,” said research lead Sridevi V. Sarma, a Johns Hopkins biomedical engineering professor. “We can get to the right diagnosis three times faster because patients often need multiple EEGs before abnormalities are detected, even if they have epilepsy. Accurate early diagnosis means a quicker path to effective treatment.”

A report of the work is published in Annals of Neurology.

Epilepsy causes recurrent, unprovoked seizures triggered by bursts of abnormal electrical activity in the brain. Standard care involves scalp EEG recordings during initial evaluations. These tests track brainwave patterns using small electrodes placed on the scalp.

Clinicians partly rely on EEGs to diagnose epilepsy and decide whether patients need anti-seizure medications. However, EEGs can be challenging to interpret because they capture noisy signals and because seizures rarely occur during the typical 20–40 minutes of an EEG recording. These characteristics make diagnosing epilepsy subjective and prone to error, even for specialists, said Sarma.

To improve reliability, Sarma’s team studied what happens in patients' brains when they are not experiencing seizures. Their EpiScalp tool uses algorithms trained on dynamic network models to map brainwave patterns and identify hidden signs of epilepsy from a single routine EEG.

“If you have epilepsy, why don’t you have seizures all the time? We hypothesised that some brain regions act as natural inhibitors, suppressing seizures. It’s like the brain’s immune response to the disease,” said Sarma.

The new study analysed 198 epilepsy patients, 91 of which had epilepsy while the rest had non-epileptic conditions mimicking epilepsy.

When Sarma’s team reanalysed the initial EEGs using EpiScalp, the tool ruled out 96 per cent of those false positives, cutting potential misdiagnoses among these cases from 54 per cent to 17 per cent.

“This is where our tool makes a difference because it can help us uncover markers of epilepsy in EEGs that appear uninformative, reducing the risk of patients being misdiagnosed and treated for a condition they don’t have,” said Khalil Husari, co-senior author and assistant professor of neurology at Johns Hopkins. “These patients experienced side effects of the anti-seizure medication without any benefit because they didn’t have epilepsy. Without the correct diagnosis, we can’t find out what’s actually causing their symptoms.”

In certain cases, misdiagnosis happens due to misinterpretation of EEGs, Husari explained, as doctors may overdiagnose epilepsy to prevent the dangers of a second seizure. In some cases, patients experience nonepileptic seizures, which mimic epilepsy. These conditions can often be treated with therapies that do not involve epilepsy medication.

MORE FROM MEDICAL & HEALTHCARE

In earlier work, the team studied epileptic brain networks using intracranial EEGs to demonstrate that the seizure onset zone is inhibited by neighbouring regions in the brain when patients are not seizing. EpiScalp builds on this research, identifying these patterns from routine scalp EEGs.

Traditional approaches to improve EEG interpretation often focus on individual signals or electrodes. Instead, EpiScalp analyses how different regions of the brain interact and influence one another through a complex network of neural pathways, said Patrick Myers, first author and doctoral student in biomedical engineering at Johns Hopkins.

The team is now conducting a larger prospective study to further validate its findings across three epilepsy centres and filed a patent for the EpiScalp technology in 2023.