Biomedical Computer Science and Mechatronics

UQ4EMEG - Uncertainty Quantification for EEG/MEG Source Analysis


Duration: 2022 - 2026

Principal Investigator (UMIT TIROL): Dr. Johannes Vorwerk
Cooperation Partner: Prof. Dr. Carsten Wolters (University of Münster, Germany), PD Dr. Stefan Rampp (University Clinic Erlangen, Germany), Prof. Dr. Sampsa Pursiainen (Aalto University, Finland), Dr. Konstantin Weise (Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany)
Funding:  Austrian Science Fund (FWF), project P35949-B
Project Description

With a prevalence of about 0.5-1.0%, epilepsy is one of the most common neurological diseases. About one-third of epilepsy patients are drug-refractory, which means they cannot be cured by treatment with anti-epileptic drugs, leaving brain surgery as the most effective treatment option.

In such a surgery, the brain area from which the epileptic seizures originate, the so-called epileptogenic zone, is removed. Unfortunately, only 15-20% of drug-refractory patients are considered eligible for surgery, often because the epileptogenic zone cannot be localized with sufficient accuracy or overlaps with important brain areas that cannot be removed. This leaves the remaining patients without a viable treatment option and the severe risks of uncontrolled seizures.

Electroencephalography (EEG) and magnetoencephalography (MEG) are tools to measure the electric and magnetic fields that arise from brain activity. To determine the active brain areas from which the measured fields originate, the use of mathematical algorithms is necessary. The reconstruction of the active brain areas from raw EEG/MEG signals is called source analysis.

EEG/MEG source analysis is an important tool to determine the epileptogenic zone in the planning of epilepsy surgery. A more accurate and reliable source localization improving the determination of the epileptogenic zone is highly valuable, as it not only has the potential to improve the results of epilepsy surgery but also to render more patients eligible for surgery.

There are different factors influencing the accuracy of EEG/MEG source analysis. One important source of uncertainty is insufficient knowledge about the electrical conductivities of the different head tissues, which differ between individuals. The exact knowledge of these conductivities is very important to accurately simulate the flow of electrical currents resulting from brain activity in the human head, which is a basic step for EEG/MEG source analysis. Unfortunately, it is almost impossible to directly measure the electrical conductivity of these tissues in (alive) subjects.

In this research project, we aim to develop novel approaches to individually estimate head tissue conductivities based on simultaneous EEG/MEG measurements. Furthermore, we will develop techniques to estimate the remaining uncertainty of the EEG/MEG source analysis, so that the EEG/MEG source analysis does not simply point to the most probable origin of the brain activity, but also shows the distribution of further possible – but less likely - source positions.

We envision that the explicit visualization of uncertainty will enable clinicians to obtain more accurate estimates of the epileptogenic zone in presurgical epilepsy diagnosis and to better judge the reliability of the performed EEG/MEG source analysis.