Recurrence Analysis for the Characterisation & Classification of Epileptic Patients

Background

Epilepsy is the most common primary neurological disorder worldwide. There is pressing need to develop better tools, investigate potential biomarkers and understand neurobiological basis of epilepsy. In clinical practice, electroencephalography (EEG) signal analysis is common and relies on visual pattern recognition of waveform abnormalities. This line of analysis respects that both epilepsy and underlying neural dynamics are highly dynamic and oscillatory phenomena. However, quantitative and automatic analysis is still under development.

The aim of the proposed project is to consolidate the methodology based on “recurrence analysis” developed over our recent GW4 Seed Corn project and apply it to data recorded from patients affected by epilepsy. Recurrence analysis is a well-established framework for nonlinear numerical time series, which allows us to (i) obtain visual information by means of two-dimensional “recurrence plots” and (ii) quantify different aspects of the dynamics by means of the so-called “recurrence quantification analysis”.

We are focusing on the GW4 grand challenge area “Health, demographic change and wellbeing”. More specifically, we are focusing on the health area, where we aim to deploy our new computational methods to improve classification and diagnosis of epilepsy.

Project summary

We proposed a new framework to assess the dynamics of brain networks based on recurrence analysis. Our framework uses recurrence plots and recurrence quantification analysis to characterize dynamic networks. For resting-state magnetoencephalographic dynamic functional networks (dFNs), we have found that functional networks recur more quickly in people with epilepsy than healthy controls. This suggests that recurrence of dFNs may be used as a biomarker of epilepsy. For stereo electroencephalography data, we have found that dFNs involved in epileptic seizures emerge before seizure onset, and recurrence analysis allows us to detect seizures. We further observe distinct dFNs before and after seizures, which may inform neurostimulation strategies to prevent seizures. Our framework can also be used for understanding dFNs in brain function in health and other neurological disorders.

This work was published in a paper and presented at a conference, and the community leads continue to collaborate even though they have all left GW4 institutes.