AI from DTU to help diagnose epilepsy

About 50 million people worldwide have epilepsy. The neurological disease causes seizures that are caused by disturbances in the electrical activity of the brain and can lead to premature death. Epilepsy can be treated with cheap medicine, but patients in low middle income countries are rarely diagnosed because of a lack of examination equipment and neurologists.

Through the eGAP project, DTU Compute will help the company BrainCapture (established in 2019 as a spinout of the research in the same department) to develop a cheap and mobile EEG scanning method based on artificial intelligence (AI), which health professionals without training can use locally in low middle income countries.

DTU develops technology for people and contributes to solving the global challenges formulated in the UN’s 17 Sustainable Development Goals. Therefore, eGAP makes particularly good sense, says associate professor in the research section Cognitive Systems at DTU Compute Tobias Andersen:

"WHO estimates that approximately 40 million people with epilepsy live in low resource settings, where up to 75% do not receive treatment. It is estimated that about 70% of them could be seizure-free if properly diagnosed and treated. Even if BrainCapture was to reach only a small proportion of these people, it would be able to improve the quality of life for millions of people."

DTU algorithm to detect disturbing data
EEG (electroencephalography) is a technique for measuring parts of brain activity. In short, a cap with electrodes is placed on the patient's head and the voltage differences between the electrodes are measured. Data is sent to a computer, where the neurologist can read them as curves and make the diagnosis based on them. Usually, a diagnosis can be made after just 20 minutes of scanning.

However, data can be difficult to interpret because eye movements, blinking and muscle contractions e.g. in the jaw muscles also give signals and mixes with the neural signals which are to show if the patient is ill.

In eGAP, data still has to be processed by a neurologist, but the actual examination is moved to local health centers, and this increases the risk of untrained staff making mistakes. DTU's algorithm based on AI must ensure the quality of the study by performing an automatic analysis of data in real-time. This will make it possible during the actual scan to detect whether the measurements are disturbed.

Unlike other similar control methods, the DTU Compute model can extract signals and map exactly which movements cause the signals. This means it will be easier to help patients not to move in a disturbing way.

"By adding some of the latest and most advanced technology in the field to BrainCapture's system, we hope to push the limits of how much the algorithms can handle on a small, inexpensive smartphone," says Tobias Andersen.

The technology must mature
The first version of the DTU algorithm is expected to be ready within a few months. When the software is installed in BrainCapture's equipment, it has to be tested by staff and patients at Filadelfia, an epilepsy hospital.

"Today, BrainCapture has simple quality control and an EGG measurement, and we can send data to a cloud platform. We now have to lift and mature the technology so that our solution can be approved as medical equipment and commercialized,” says CEO of BrainCapture and project manager Tue Lehn-Schiøler.

It is expected that the eGAP method will be able to map approximately 60% of all epilepsy cases and thus help many patients. Initially, BrainCapture concentrates on Kenya, where it has partners.