A novel tool for the removal of muscle artefacts from EEG: Improving data quality in the gamma frequency range

Investigation on how dynamic effective connectivity patterns encode the fluctuating pain intensity in chronic migraine

Chronic migraine is characterised by persistent headaches for >15 days per month; the intensity of the pain is fluctuating over time. Here, we explored the dynamic interplay of connectivity patterns between regions known to be related to pain processing and their relation to the ongoing dynamic pain experience. We recorded EEG from 80 sessions (20 chronic migraine patients in 4 separate sessions of 25 min). The patients were asked to continuously rate the intensity of their endogenous headache. On different time-windows, a dynamic causal model (DCM) of cross spectral responses was inverted to estimate connectivity strengths. For each patient and session, the evolving dynamics of effective connectivity were related to pain intensities and to pain intensity changes by using a Bayesian linear model. Hierarchical Bayesian modelling was further used to examine which connectivity-pain relations are consistent across sessions and across patients.

The results reflect the multi-facetted clinical picture of the disease. Across all sessions, each patient with chronic migraine exhibited a distinct pattern of pain intensity-related cortical connectivity. The diversity of the individual findings are accompanied by inconsistent relations between the connectivity parameters and pain intensity or pain intensity changes at group level. This suggests a rejection of the idea of a common neuronal core problem for chronic migraine.

Ensemble Deep Neural Network for Automatic Classification of EEG Independent Components

OBJECTIVE

Independent component analysis (ICA) is commonly used to remove noisy artifacts from multi-channel scalp electroencephalogram (EEG) signals. ICA decomposes EEG into different independent components (ICs) and then, experts remove the noisy ones. This process is highly time-consuming and experts are not always available. To surpass this drawback, research is going on to develop models to automatically conduct IC classification. Current state-of-the-art models use power spectrum densities (PSDs) and topoplots to classify ICs. The performance of these methods may be limited by disregarding the IC time-series that would contribute to fully simulate the visual inspection performed by experts.

METHODS

We present a novel ensemble deep neural network that combines time-series, PSDs, and topoplots to classify ICs. Moreover, we study the ability to use our model in transfer learning approaches.

RESULTS

Experimental results showed that using time-series improves IC classification. Results also indicated that transfer learning obtained higher performance than simply training a new model from scratch.

CONCLUSION

Researchers should develop IC classifiers using the three sources of information. Moreover, transfer learning approaches should be considered when producing new deep learning models.

SIGNIFICANCE

This work improves IC classification, enhancing the automatic removal of EEG artifacts. Additionally, since labelled ICs are generally not publicly available, the possibility of using our model in transfer learning studies may motivate other researchers to develop their own classifiers.