Final Report Summary - 4D-EEG (4D-EEG: A new tool to investigate the spatial and temporal activity patterns in the brain.)
In the 4D EEG project a new methodology was developed to monitor brain activity with a very high spatial (3D) and temporal (+1D = 4D) resolution (target: 2 mm and 1 msec). In comparison, fMRI records the brain activity with a high spatial resolution (2 mm) and is based on oxygen consumption which has a time lag of seconds (!).
Fundamental control engineering theory dictates that a functioning control loop can only be measured by imposing well-known perturbations and observe the response at various locations in the control loop. Our unique methodology is based on the use of robot manipulators to impose force or motion perturbations to the wrist, and measure the resulting forces, positions, electrical signals at the muscles (EMG) and in the brain (EEG).
We have designed the perturbation signals in the frequency domain as a multi-sine signal, which frequencies we also retrieved in the EEG signal. This approach resulted in a much better Signal to Noise Ratio (SNR) in EEG recordings than the existing techniques. An unforeseen result was that more than 80% of the EEG signal power was due to non-linearities in the system, as shown as higher harmonics of the perturbation frequencies in the EEG signal. Most existing EEG methodology is based on linear assumptions. We were forced to adapt our initial linear approach and developed new mathematical methods to characterize the non-linearities. We have developed several approaches for non-linear dynamic modelling , i.e. Multi-Spectral Phase Coherence (MSPC) method, Cross-frequency Amplitude Transfer Function (CATF) and Volterra models, which are typically capable of modelling about 50% of the non-linear signal content.
High-density EEG recordings enabled us to estimate the location of EEG sources, i.e. the active brain areas. We have improved the EEG source localisation by prewhitening the EEG signal and using the co-variance to improve the SNR. Applying Independent Component Analysis (ICA) techniques we could also investigate which sources mostly contributed to the EEG signal.
Diffusion Tensor Imaging (DTI) is a MRI technique to visualize the anatomical structure of the pathways through the brain. We have combined DTI with functional 4D EEG measurements to analyse the anatomical and physiological connectivity between brain areas. The first linear dynamic model showed how the signal propagates through the cortical areas, as a response to a peripheral stimulus. Nonlinear modelling approaches are under development.
Our first medical application is in patients with a stroke (CerebroVascular Accident, CVA). Stroke patients with a lesion in the sensorimotor cortex suffer from functional impairments in the contralesional limbs. In case of a severe stroke, the other half of the brain may partly take over the lost function. Functional recovery can occur through restitution (retrieval of pre-existing function at the lesion site) or compensation (other brain regions taking over functions). The 4D EEG project enabled monitoring of the recovery process in time, and optimization of rehabilitation therapy.
We have performed a cross-sectional study with chronic stroke patients to establish the experimental set-up, which is comprised of EEG resting state, Somatosensory Evoked Potentials (SEP) and force perturbations. We have applied the 4D EEG methodology in a longitudinal study with acute stroke patients, recording within one week after the incident, and at 5, 12 and 26 weeks follow-up. For the patient recordings we have built a mobile EEG laboratory in a van, so that we could visit the patients at their home, or in the rehabilitation or nursing clinic. The longitudinal study resulted in a unique prognostic model about the functional recovery which will improve the therapy prescription. Most important parameters in the prognostic model were the lateralization of the cortical activity, i.e. to what extent the opposite site of the lesion might take over the function, and the connectivity of the sensory signals to the lesion side of the brain.
The 4D EEG methodology gives us an unprecedented opportunity for fundamental analysis of role of the brain in human motor control, and has also been used with the visual system. 4D EEG is a non-invasive method which can be applied during functional tasks (e.g. airplane control, car driving, object manipulation).
Fundamental control engineering theory dictates that a functioning control loop can only be measured by imposing well-known perturbations and observe the response at various locations in the control loop. Our unique methodology is based on the use of robot manipulators to impose force or motion perturbations to the wrist, and measure the resulting forces, positions, electrical signals at the muscles (EMG) and in the brain (EEG).
We have designed the perturbation signals in the frequency domain as a multi-sine signal, which frequencies we also retrieved in the EEG signal. This approach resulted in a much better Signal to Noise Ratio (SNR) in EEG recordings than the existing techniques. An unforeseen result was that more than 80% of the EEG signal power was due to non-linearities in the system, as shown as higher harmonics of the perturbation frequencies in the EEG signal. Most existing EEG methodology is based on linear assumptions. We were forced to adapt our initial linear approach and developed new mathematical methods to characterize the non-linearities. We have developed several approaches for non-linear dynamic modelling , i.e. Multi-Spectral Phase Coherence (MSPC) method, Cross-frequency Amplitude Transfer Function (CATF) and Volterra models, which are typically capable of modelling about 50% of the non-linear signal content.
High-density EEG recordings enabled us to estimate the location of EEG sources, i.e. the active brain areas. We have improved the EEG source localisation by prewhitening the EEG signal and using the co-variance to improve the SNR. Applying Independent Component Analysis (ICA) techniques we could also investigate which sources mostly contributed to the EEG signal.
Diffusion Tensor Imaging (DTI) is a MRI technique to visualize the anatomical structure of the pathways through the brain. We have combined DTI with functional 4D EEG measurements to analyse the anatomical and physiological connectivity between brain areas. The first linear dynamic model showed how the signal propagates through the cortical areas, as a response to a peripheral stimulus. Nonlinear modelling approaches are under development.
Our first medical application is in patients with a stroke (CerebroVascular Accident, CVA). Stroke patients with a lesion in the sensorimotor cortex suffer from functional impairments in the contralesional limbs. In case of a severe stroke, the other half of the brain may partly take over the lost function. Functional recovery can occur through restitution (retrieval of pre-existing function at the lesion site) or compensation (other brain regions taking over functions). The 4D EEG project enabled monitoring of the recovery process in time, and optimization of rehabilitation therapy.
We have performed a cross-sectional study with chronic stroke patients to establish the experimental set-up, which is comprised of EEG resting state, Somatosensory Evoked Potentials (SEP) and force perturbations. We have applied the 4D EEG methodology in a longitudinal study with acute stroke patients, recording within one week after the incident, and at 5, 12 and 26 weeks follow-up. For the patient recordings we have built a mobile EEG laboratory in a van, so that we could visit the patients at their home, or in the rehabilitation or nursing clinic. The longitudinal study resulted in a unique prognostic model about the functional recovery which will improve the therapy prescription. Most important parameters in the prognostic model were the lateralization of the cortical activity, i.e. to what extent the opposite site of the lesion might take over the function, and the connectivity of the sensory signals to the lesion side of the brain.
The 4D EEG methodology gives us an unprecedented opportunity for fundamental analysis of role of the brain in human motor control, and has also been used with the visual system. 4D EEG is a non-invasive method which can be applied during functional tasks (e.g. airplane control, car driving, object manipulation).