Electroencephalography (EEG) is the recording of electrical activity along the scalp, measuring voltage fluctuations resulting from current flows within the neurons of the brain. Being a relatively low cost non-invasive method EEG has an important role in diagnosis and monitoring of a long list of neurological conditions (with epilepsy being a bold example), and is a valuable tool for brain research. One of the foremost advantages of EEG compared to other brain imaging modalities is its high temporal resolution (on the order of milliseconds). The downside of EEG is, however, its poor spatial resolution. We propose a novel concept of EEG measurement hardware which, in combination with signal processing techniques, will increase both the angular and the depth resolution of EEG. Our idea is based on the observation that by connecting a network of controllable impedances between pairs of measurement electrodes and allowing current to flow between them, one can control the shape of the spatial filter constituted by the skull. In addition to introducing many independent equations to the EEG inverse problem, these new measurements will also allow non-invasive estimation of the electrical conductivity of the head thus improving the quality of the forward model. We expect these two contributions alone to improve significantly the spatial resolution of EEG source estimation that relies on the solution of the inverse problem. Better source analysis would improve additional emerging applications of EEG such as neuro-feedback and brain-computer interface (BCI).
To test our idea we developed the phantom mimicking the electromagnetic properties of the human head. The novel key features of the phantom we developed using the 3d-printing technology is the realistic anisotropic electrical conductivity of the skull and the possibility to place sufficiently dense current sources. We also developed a system for controllable local heating of the conductive gel, which in turn enables local controllable change in the conductivity of the medium suitable for electrical impedance tomography studies. We believe the phantom we developed will be useful to the broad neuroscience community.
The simulation studies we performed corroborate our hypothesis. However, we found the method too sensitive to the quality of the electrode-skin contact impedance which for most practical cases is unstable especially for the consumable electronics applications. Solution of these problems led us to the following two complementing methods:
1) The patent pending method which by employing mechanical vibration of cerebral cortex is able to increase significantly the precision of the EEG inverse problem and is robust enough to the model parameters due to the feedback provided by the vibration monitoring. This method has the potential to be for the clinical applications and neuroscience. Our simulation studies were followed by the 3d-printing of the flexible cerebral cortex for the subsequent testing of the method.
2) The patent pending method in which a capacitive coupling of the electrophysiological signals to the electrodes, an impedance network and current sources are employed. Using algorithms from the compressed sensing and machine learning families the new information that is not contained in the regular measurements is extracted. Processing of this information increases the robustness of wearable devices such as gesture recognition bracelets or VR helmets to the movement artifact from the one hand and reduces the number of required measuring channels from the other hand increasing quality/cost ratio of the wearables. This method also reduces the power consumption of wireless wearables leading to longer battery life. Another use of this invention can be for medical devices. For instance it can reduce the costs of the dense EEG systems (comprising many electrodes) and make it affordable for any clinic.
We also made preliminary experiments on the phantom towards the testing of the NetEEG method for invasive devices. We found that this method could be especially beneficial for the epileptic patients having depth electrodes since for these electrodes the contact impedance is low and stable.
Based on our simulation studies we developed and manufactured a circuit implementing the load modulation of the measured electrophysiological signals. The board is still being fully characterized at the time of this report.
The mechanical models of the dense EMG bracelet prototype for the BCI applications and the EEG helmet for the clinical and BCI applications suitable for 3d-printing were developed. The helmet we developed is a patent pending technology for reducing pressure injuries by long-time worn dry electrodes by epileptic patients.
