Developing advanced graph filtering methods and models for robust assessment of transient EEG functional connectivity

This project will explore links between Graph Neural Networks (GNN) and functional connectivity of the animal brain with applications to EEG functional connectivity including for early detection of Alzheimer's disease.

Literature has clearly established that it is neither spatial nor temporal localization of brain activity that causes cognitive phenomena and the corresponding brain function but in fact, how the different areas of the brain are dynamically interconnected over the evolution of time. The electroencephalogram (EEG) contains important discriminating information relating to sequential brain process in response to various cognitive tasks.

Providing a very high temporal resolution, scalp EEG allows for the direct recording of electromagnetic activity of the brain in a non-invasive relatively cheap way. Scalp EEG presents several notable limitations however, with the most prominent being the substantial noise levels inherent in the recorded signals. This noise poses a significant challenge, especially when attempting to investigate the functional connectivity associated with transient cognitive processes occurring within mere tens of milliseconds.

A pivotal issue within the realm of EEG signal analysis pertains to the extraction of dependable connectivity estimates with this temporal precision. Measuring dynamic functional brain connectivity in short time windows is gaining increasing recognition in AD research due its potential to provide information for the early detection of the devastating disease.

An important reason for this is the growing recognition that intricate changes in brain connectivity can occur before the onset of clinical symptoms, this makes it a promising avenue for early bio-marker development and better understanding of disease progression.

The first aim of the project is to provide advanced tools for the robust analysis of short-term functional connectivity from the Electroencephalogram (EEG). This will illuminate transient activity of the human brain at the level of tens of milliseconds and work towards sensitive and specific detection of dementia from EEG signals.

The second aim is to develop advanced predictive machine learning models which can exploit these methods. We will develop and implement novel machine learning models for use in the prediction of activity from both resting state and cognitive tasks. For example, we will consider the development of a specific framework for Machine learning using graph variate signal analysis [1]- 'Graph Variate Neural Networks'.

The goal here would be to overcome existing issues in dynamic functional connectivity analysis in small temporal windows.

The third aim is to apply these methods towards the sensitive and specific detection of Alzheimer's disease and other dementias at early stages to ameliorate the socio-economic burdens associated with these illnesses. Using data from collaborators within the University (the EURO-LAD consortium containing EEG data collected from over 500 participants with various forms of dementia including Alzheimer's disease, vascular dementia and frontotemporal dementia), we will apply these methods to the problem of detecting early-stage Alzheimer's disease using EEG.

[1] Smith KM, Spyrou L, Escudero J, IEEE Transactions on Signal Processing, 67(2): 293-305 (2019)