A team of researchers from HSE University and the Artificial Intelligence Research Institute (AIRI) have demonstrated the effectiveness of the PSIICOS method they had previously developed for non-invasive mapping the neural networks in the brain based on its electrical activity. Unlike other methods, it does not search for individual neuronal sources to be then combined into networks but instead looks directly for the functional networks of interconnected neuronal populations—and does so swiftly and accurately. The study findings have been published in NeuroImage.
According to a recent theory, the brain is a 'hyper network' composed of numerous smaller neural networks. These smaller networks can synchronise to perform various functions; thus, they are referred to as functional networks. For instance, when a person encounters an aggressive dog, the visual network collaborates with the motor control network, leading the individual to react by running away. Disruptions in functional connectivity in the brain can contribute to the onset of neurological disorders. Research has demonstrated that certain neurological pathologies are caused by disruptions in connectivity between various brain regions rather than by issues within a particular area.
Electroencephalography (EEG) and magnetoencephalography (MEG), often employed in functional neuroimaging studies, allow for the conventional determination of the location and temporal profiles of the hubs of electrical activity in the brain, facilitating the subsequent detection of their functional connectivity using a range of mathematical methods. However, these methods do not enable the detection of functional networks with perfectly synchronous nodes. In 2018, the team of scientists from the HSE Centre for Bioelectric Interfaces introduced, for the first time, a mathematical algorithm capable of detecting such networks through EEG and MEG measurements of brain activity. At the core of the new approach is the principle of searching specifically for functional networks rather than for individual neuronal sources.
In a new study, researchers from HSE University and AIRI have demonstrated that the mathematical method they introduced earlier offers the highest accuracy without significant additional computational costs. This approach is founded on the PSIICOS projection, a mathematical operation that aids in filtering out the brain's background activity while focusing on functional networks. The PSIICOS method proves particularly effective when dealing with strictly synchronised (no delay) activation of neural sources, and enables the detection of a functional connection between them.
Picture yourself standing in a crowd with your eyes closed, surrounded by people who are all singing. Each of them is singing a different song, except for two who are singing the same melody in unison. This is what you need to focus on. Our method enables us to disregard solo singers as much as possible, locate the couple singing in unison, zoom in on it, and identify the melody this couple is singing. Furthermore, if multiple couples are singing in unison, our approach will enable the detection and identification of each duo and their respective melodies. It is fundamentally significant that we aim to directly detect couples singing in unison, while other methods typically involve two steps: first listening to each individual singer, and then pairing them one by one to identify the couple that sings synchronously and in unison.
Building upon the PSIICOS method, the researchers intend to create several novel solutions for non-invasive mapping of functional networks, regardless of the time lag between the temporal activity profiles of neural populations in the network nodes. In the future, the method may also be applied using EEG, which is a more accessible method of neuroimaging than MEG.
According to the scientists, the ability to monitor the operation of functional brain networks both during cognitive or motor tasks and at rest will enable the development of objective tests for diagnosing and predicting neurodegenerative disorders.