In the last decades, several studies have been carried on in order to extract the brain networks at the basis of mental processes, which are characterized by lots of interactions between different and differently specialized cortical and subcortical sites. Electroencephalography (EEG) is widely used in the study of the directed functional connectivity due to its high temporal resolution and low invasiveness but the meaning of the directed links detected at scalp level is not clear. The main open issue is related with the uncertainty introduced by two main evidences: i) the spread of the electrical field generated by neural activity and ii) the impossibility to associate one generator (source) to one sensors. Previous studies have already demonstrated the presence of a bias affecting all the brain connectivity measures related with the presence of the head volume conduction for different estimators and algorithms for the source reconstruction. A complete characterization and quantification of such investigated problem in different controlled experimental conditions in still missing. The first goal of the present project is to analyze and quantify the effects of the volume conduction on the brain connectivity estimates and to provide reliable solutions in order to increase their accuracy and their physiological interpretability. It will allow to increase the accuracy of widely used connectivity estimators and their value as neuro-physiological descriptors. The second aim of the work will be the extraction of synthetic connectivity-based measures able to catch the same local and global properties on the scalp and source level networks. Obtained results could lead to evidences proving the possibility to give a correct interpretation of the networks directly analyzed in the sensors domain.
The results of the present project could have a wide impact. First of all, we will develop a toolbox able to: i) generate EEG simulated data with any type of characteristic required from the investigated phenomena and with a well-known connectivity pattern; ii) to impose a network with controlled topographical properties. We will use it to build the two proposed simulation study but it could have a large and more general impact allowing to quantify each bias that a user wants to consider and to test current and new methodologies to overcome these biases.
The main advancement related with the results of the study will concern the meaning and the physiological interpretation of connectivity measures at source and scalp level. On the other side we¿ll quantify the estimation error in different experimental conditions providing some guidelines that will allow to increase the accuracy of widely used connectivity estimates and their value as neuro-physiological descriptors.
In conclusion, the innovations of the project will cover several aspects:
- methodological, producing a quantification of the errors on the connectivity estimates due to the volume conduction effect and a series of guidelines that will allow to take into account the inaccuracy of the results under specific conditions.
- neuro-physiological, allowing to give a more reliable interpretation to the networks extracted at source level and to the measure extracted at scalp level.
- clinical, identifying a set of reliable synthetic measures able to well describe the main properties of the real brain networks through the simply analysis of the scalp EEG signals. They could be used in several clinical application also during real time application (for example neuro-feedback training for the recovery of a cognitive function) or when subjects with brain lesion are involved.