According to the World Health Organization, epilepsy affects nearly 50 million people worldwide, and about 70% of them living with epilepsy could live seizure-free if properly diagnosed and treated. However, only manual inspection of electroencephalogram (EEG) reports by skilled neurologists is available for early epilepsy detection.
To alleviate this issue, researchers at the Indian Institute of Science (IISc), in collaboration with physicians at the All India Institute Of Medical Science (AIIMS) Rishikesh, have developed a novel algorithm to detect epilepsy from EEG data.
If you or someone you love has ever suffered a seizure, this new algorithm has the potential to help you live a normal life with epilepsy. Let’s know more about it.
What is epilepsy?
Epilepsy is a neurological disease categorised by abnormal activity in the brain. In epilepsy, the brain suddenly emits electrical signals, similar to electrical storms, resulting in seizures, fits and, in extreme cases, death.
Epilepsy is characterised by two types of recurrent seizures—focal and generalised—based on the location of occurrence. Focal epilepsy is when they occur in a specified range and generalised when they occur at random locations.
Early detection of epilepsy is essential, as in some cases, epilepsy can lead to death within a few hours if appropriate treatment isn’t provided. While several treatment options like medications and responsive stimulation devices are available for epilepsy, only manual inspection of electroencephalogram (EEG) reports by skilled neurologists is available for early detection.
However, manual EEG inspections are time-consuming, labour-intensive and fairly subjective, resulting in misclassification. This propelled researchers to develop an algorithm that can automatically screen and diagnose epilepsy based on EEG signals.
How does the algorithm to detect epilepsy work?
In comparison to a healthy person, an epileptic patient shows a different set of brainwave patterns. The algorithm developed by IISc researchers utilises machine learning and diagnostic techniques to distinguish between normal and epileptic EEGs.
“The research aims to differentiate EEG of normal subjects from epileptic EEGs. Additionally, the developed algorithm attempts to identify the types of seizures. Our work is to help the neurologists make an efficient and quick automated screening and diagnosis,”
– Hardik J Pandya, Assistant Professor at the Department of Electronic Systems Engineering (DESE) and the corresponding author of the study published in Biomedical Signal Processing and Control.
Brain waves recorded during EEGs can be classified into different wave patterns—sharp signals, spikes and slow waves. The algorithm calculates the total number of sharp waves—cumulative sharp count—and uses this parameter to classify EEGs. A higher value indicates epileptic EEG. It can also distinguish between focal and generalised epilepsy by calculating the sum of areas under spike and sharp curves.
Furthermore, the algorithm could also identify absence seizures, which are characterised by a brief and sudden loss of consciousness, using a cumulative spike wave count. The absence seizures can be critical and fatal, making early diagnosis essential.
Algorithm training and development
To train, the algorithm researchers collected EEG data of 88 epileptic patients from AIIMS Rishikesh, analysed and classified brainwave patterns and fed it to the algorithm. The algorithm then correctly classified EEGs into two categories with a high degree of accuracy.
To check algorithm accuracy, the researchers ran it on new sets of EEG data. The doctors already knew whether patients had epilepsy and, if so, what type of epilepsy. The researchers say this blind validation study correctly classified the EEGs in nearly 91% of the cases.
Parting Words
While the new algorithm is under research and development, its underlying potential is yet to be discovered. Considering the numerous positive reviews it has received, there is a high chance it can be used for emergency epilepsy and sudden stroke detection in the coming years.
Currently, the research is published in the Biomedical Signal Processing and Control for peer review, the algorithm is being tested for its reliability by physicians at AIIMS Rishikesh, and a patent for the work has been filed by IISc researchers. They have taken the first step towards epilepsy semiology and management; many more are yet to come.
“We hope to refine this further by testing on more data to consider more variabilities of human EEGs until we reach the point where this becomes completely translational and robust,”
– Rathin K Joshi, a PhD student in DESE and first author of the study.
