Volume Conduction Problem Research Articles

The effects of spatial leakage correction on the reliability of EEG-based functional connectivity networks.

Electroencephalography (EEG) functional connectivity (FC) estimates are confounded by the volume conduction problem. This effect can be greatly reduced by applying FC measures insensitive to instantaneous, zero-lag dependencies (corrected measures). However, numerous studies showed that FC measures sensitive to volume conduction (uncorrected measures) exhibit higher reliability and higher subject-level identifiability. We tested how source reconstruction contributed to the reliability difference of EEG FC measures on a large (n = 201) resting-state data set testing eight FC measures (including corrected and uncorrected measures). We showed that the high reliability of uncorrected FC measures in resting state partly stems from source reconstruction: idiosyncratic noise patterns define a baseline resting-state functional network that explains a significant portion of the reliability of uncorrected FC measures. This effect remained valid for template head model-based, as well as individual head model-based source reconstruction. Based on our findings we made suggestions how to best use spatial leakage corrected and uncorrected FC measures depending on the main goals of the study.

EEG source imaging of hand movement-related areas: an evaluation of the reconstruction and classification accuracy with optimized channels

The hand motor activity can be identified and converted into commands for controlling machines through a brain-computer interface (BCI) system. Electroencephalography (EEG) based BCI systems employ electrodes to measure the electrical brain activity projected at the scalp and discern patterns. However, the volume conduction problem attenuates the electric potential from the brain to the scalp and introduces spatial mixing to the signals. EEG source imaging (ESI) techniques can be applied to alleviate these issues and enhance the spatial segregation of information. Despite this potential solution, the use of ESI has not been extensively applied in BCI systems, largely due to accuracy concerns over reconstruction accuracy when using low-density EEG (ldEEG), which is commonly used in BCIs. To overcome these accuracy issues in low channel counts, recent studies have proposed reducing the number of EEG channels based on optimized channel selection. This work presents an evaluation of the spatial and temporal accuracy of ESI when applying optimized channel selection towards ldEEG number of channels. For this, a simulation study of source activity related to hand movement has been performed using as a starting point an EEG system with 339 channels. The results obtained after optimization show that the activity in the concerned areas can be retrieved with a spatial accuracy of 3.99, 10.69, and 14.29 mm (localization error) when using 32, 16, and 8 channel counts respectively. In addition, the use of optimally selected electrodes has been validated in a motor imagery classification task, obtaining a higher classification performance when using 16 optimally selected channels than 32 typical electrode distributions under 10–10 system, and obtaining higher classification performance when combining ESI methods with the optimal selected channels.

Recognition of emotional states using frequency effective connectivity maps through transfer learning approach from electroencephalogram signals

Convolutional Neural Networks (CNN) have been widely utilized in Emotion Recognition (ER) research due to their vast benefits. Designing specific CNN configuration and learning accurate parameters are provided through the Transfer Learning (TL) approach. A novel combined ER schema based on sophisticated Frequency Effective Connectivity (FEC) maps and transfer learned CNN models named FEC-CNN is proposed in this study. The EC measure estimates the information flow between brain channels in frequency bands regarding the volume conduction problem. Then, visualization of this information provides valuable and distinctive information for classifying emotional states. After preprocessing Electroencephalogram (EEG) signals, the Effective Connectivity maps were computed through the Partial Directed Coherence (PDC) measure from EEG channels at standard delta, theta, alpha, beta, and gamma frequency bands. Then, PDC maps were used to fine-tune six pre-trained CNN models, AlexNet, ResNet-18, DarkNet-19, ShuffleNet, Inception-v3, and Xception, and classify five emotional states toward the subject independent Leave-One-Subject-Out (LOSO) Cross-Validation criterion for MAHNOB-HCI, DEAP and DREAMER databases. The experimental results show that ResNet-18 achieved the highest average accuracy and F-score 94.27%, 94.74% for the DEAP database, and 95.25% and 95.33% for the MAHNOB-HCI database, and 96.00% and 96.77% for DREAMER database, to recognize five emotional states from PDC maps at alpha frequency bands. Comparison of the experimental results with recent researches showed the superiority of the proposed EC-based CNN model.This research introduced an interesting sight of deep learning applications in cognitive computation area by utilizing effective connectivity information flows to recognize emotional states from EEG signals.

Multi-Objective Optimization-Based High-Pass Spatial Filtering for SSVEP-Based Brain–Computer Interfaces

Many spatial filtering methods have been proposed to enhance the target identification performance for the steady-state visual evoked potential (SSVEP)-based brain–computer interface (BCI). The existing approaches tend to learn spatial filter parameters of a certain target using only the training data from the same stimulus, and they rarely consider the information from other stimuli or the volume conduction problem during the training process. In this article, we propose a novel multi-objective optimization-based high-pass spatial filtering method to improve the SSVEP detection accuracy and robustness. The filters are derived via maximizing the correlation between the training signal and the individual template from the same target whilst minimizing the correlation between the signal from other targets and the template. The optimization will also be subject to the constraint that the sum of filter elements is zero. The evaluation study on two self-collected SSVEP datasets (including 12 and four frequencies, respectively) shows that the proposed method outperformed the compared methods such as canonical correlation analysis (CCA), multiset CCA (MsetCCA), sum of squared correlations (SSCOR), and task-related component analysis (TRCA). The proposed method was also verified on a public 40-class SSVEP benchmark dataset recorded from 35 subjects. The experimental results have demonstrated the effectiveness of the proposed approach for enhancing the SSVEP detection performance.

Schizophrenia Classification Using Resting State EEG Functional Connectivity: Source Level Outperforms Sensor Level.

Disrupted functional and structural connectivity measures have been used to distinguish schizophrenia patients from healthy controls. Classification methods based on functional connectivity derived from EEG signals are limited by the volume conduction problem. Recorded time series at scalp electrodes capture a mixture of common sources signals, resulting in spurious connections. We have transformed sensor level resting state EEG times series to source level EEG signals utilizing a source reconstruction method. Functional connectivity networks were calculated by computing phase lag values between brain regions at both the sensor and source level. Brain complex network analysis was used to extract features and the best features were selected by a feature selection method. A logistic regression classifier was used to distinguish schizophrenia patients from healthy controls at five different frequency bands. The best classifier performance was based on connectivity measures derived from the source space and the theta band.The transformation of scalp EEG signals to source signals combined with functional connectivity analysis may provide superior features for machine learning applications.

Temporal-Spatial Patterns in Dynamic Functional Brain Network for Self-Paced Hand Movement.

Dynamic functional connectivity is attracting a growing interest as it has been suggested to be a more accurate representation of functional brain networks compared to traditional functional connectivity. It is believed that the functional connectivity fluctuations result from the transitions among different brain states other than continuous changes in the brain. In this paper, we aim to investigate the spatial-temporal changes in the interactions between different brain regions during a self-paced hand movement with EEG signals. A systematic analysis framework, consisting of connectivity metric calculation, brain state segmentation, temporal representative graph extraction, and spatial community detection, is proposed to analyze the dynamic functional connectivity. First, corrected imaginary coherency is applied to measure the functional connectivity as it is insensitive to EEG volume conduction problem. Second, singular value decomposition (SVD) vector space distance between the connectivity matrices at two adjacent time points is calculated. In addition, the brain states are segmented based on the changes in the time series of SVD vector space distances. Third, one representative graph is summarized within each state segment using the SVD vectors corresponding to the k largest singular values. Finally, spatial patterns on the representative graph are detected with a modularity-based community detection method. Based on the SVD vector space distance using the change point detection method, a series of brain states lasting for hundreds of milliseconds are identified. Moreover, we find that the sudden decrease points in SVD vector space distance coincide with early Bereitschafts potential. In addition, we find that there are several connectivity patterns along the time before the onset of movement. At first, the functional connectivity is relatively dispersed. Gradually, the functional connectivity begins to concentrate and the predominant communities in each dynamic functional network can be observed clearly.

A Tutorial Review of Functional Connectivity Analysis Methods and Their Interpretational Pitfalls.

Oscillatory neuronal activity may provide a mechanism for dynamic network coordination. Rhythmic neuronal interactions can be quantified using multiple metrics, each with their own advantages and disadvantages. This tutorial will review and summarize current analysis methods used in the field of invasive and non-invasive electrophysiology to study the dynamic connections between neuronal populations. First, we review metrics for functional connectivity, including coherence, phase synchronization, phase-slope index, and Granger causality, with the specific aim to provide an intuition for how these metrics work, as well as their quantitative definition. Next, we highlight a number of interpretational caveats and common pitfalls that can arise when performing functional connectivity analysis, including the common reference problem, the signal to noise ratio problem, the volume conduction problem, the common input problem, and the sample size bias problem. These pitfalls will be illustrated by presenting a set of MATLAB-scripts, which can be executed by the reader to simulate each of these potential problems. We discuss how these issues can be addressed using current methods.

Postoperative delirium: disconnecting the network?

Postoperative delirium: disconnecting the network?

Simulations of magnetocardiographic signals using realistic geometry models of the heart and torso

Abstract. Although the first measurement of the cardiac magnetic field was reported almost half a century ago magnetocardiography (MCG) is not yet widely used as a clinical diagnostic technique. With the development of a new generation of magnetoelectric sensors it is believed that MCG will become widely accepted in the clinical diagnosis. Our goal is to build a computer-based tool for medical diagnosis and to use it for the clarification of open electro-physiological questions. Here we present results from modelling of the cardiac electrical activity and computation of the generated magnetic field. For the simulations we use MRT-based anatomical models of the human atria and ventricles where the shape of the action potential is determined by ionic currents passing through the cardiac cell membranes. The monodomain reaction-diffusion equation is chosen for the description of the heart's electrical activity. This equation is solved for the transmembrane voltage which is in turn used to calculate current densities at discrete time instants. In subsequent simulations these current densities represent primary sources of magnetostatic fields arising from a volume conduction problem. In these simulations the heart is placed in a realistic torso model where the lungs are also considered. Both, the volume conduction problem as well as the reaction-diffusion problem are modelled using Finite-Element techniques.

Volume conduction effects in brain network inference from electroencephalographic recordings using phase lag index

In this paper, we test the performance of a synchronicity estimator widely applied in Neuroscience, phase lag index (PLI), for brain network inference in EEG. We implement the four sphere head model to simulate the volume conduction problem present in EEG recordings and measure the activity at the scalp of surrogate sources located at the brain level. Then, networks are estimated under the null hypothesis (independent sources) using PLI, coherence (R) and phase coherence (PC) for the volume conduction and no volume conduction (NVC) cases. It is known that R and PC are highly influenced by volume conduction, leading to the inference of clustered grid networks. PLI was designed to solve this problem. Our simulations show that PLI is partially invariant to volume conduction. The networks found by PLI show small-worldness, with a clustering coefficient higher than random networks. On the contrary, PLI-NVC obtains networks whose distribution is closer to random networks indicating that the high clustering shown by PLI networks are caused by volume conduction. The influence of volume conduction in PLI might lead to biased results in brain network inference from EEG if this behaviour is ignored.

3D mesh generation to solve the electrical volume conduction problem in the implanted inner ear

Cochlear implants are used to restore hearing in the profoundly deaf [Th.J. Balkany, Otolaryngol. Clin. North. Am. 19 (2) (1986) 215–449] by direct electrical stimulation of the auditory nerve. To study the working mechanism of cochlear implants and to provide a tool to develop better ones, a Boundary Element electrical volume conduction model of the cochlea (the auditory part of the inner ear) has been constructed. In this paper first a short comparison of the available numerical methods is given, then an algorithm is presented with which different cochlear geometries can be constructed and fitted with different types of cochlear implants. With the resulting model the potential distributions induced by the implant can be calculated, and a prediction of the effect of the implant can be made. The use of the meshing algorithm is not restricted to cochlear implants, but is also applicable in other fields.

Numerical Calculation of the Potential Distribution Due to Dipole Sources in a Spherical Model of the Head

A three-dimensional spherical model of the head was investigated numerically. The model consists of four conductive layers representing the scalp, the skull, the cerebrospinal fluid, and the cortex with a dipole current source. The potential created by the dipole was calculated using quasistatic formulation and a linear medium. The volume conduction equation was discretized by the finite volume method to ensure the conservation of fluxes and efficient solution method. The large set of algebraic equations for the electric potential was solved iteratively by the successive over relaxation method. The new formulation of the volume conduction problem was validated by comparing the numerical results with two analytical solutions. The first test-case considers a homogeneous spherical model with a dipole in the center. The potential on the outer surface, as well as within the volume conductor, was calculated and very good agreement was obtained with the analytical solution. In the second test-case, the scalp potential due to a radially oriented eccentric dipole in a four concentric spheres model was compared with an analytic solution. It was found that a grid of 90 × 90 × 90 volume elements yielded accurate results on the scalp surface with errors on the order of 1%. The present numerical model can be extended to general cases with any volume conductor shape or with any distribution or orientation of the current dipoles. Compared to other numerical methods, this approach offers enhanced accuracy for given computational resources (both in CPU time and memory). The gain might be more than one order of magnitude, allowing simulation with considerably larger meshes.

Electrical impedance tomography. Solving anisotropic mixed boundary value problems by means of the boundary element method.

In a previous study we have shown that an anisotropic volume conduction problem involving the Neumann condition at the outer surface of a volume conductor can be solved by means of the boundary element method by applying an appropriate coordinate transformation (Zhou and van Oosterom 1991). In this paper we will show that this approach can also be applied to solve the anisotropic mixed boundary value condition problems studied in electrical impedance tomography (EIT). Results are presented of numerical computations, which were carried out for a volume conductor of irregular (torso) shape.

On the numerical accuracy of the boundary element method.

The numerical accuracy of the boundary element (BE) method used to solve the volume conduction problem of nested compartments, each having a homogeneous conductivity, is studied. The following techniques for improving this accuracy are discussed: the handling of the auto solid angle element omega ii, the overall refinement of the level of discreteness, the use of a locally refined discrete grid, the isolated problem approach, and an adaptive refined computation of the discrete surface integrals involved in the BE method. The effects of these techniques on the numerical accuracy of the computed electrical potentials are illustrated by taking a volume conductor consisting of four concentric spheres representing the head since for this model an analytical (exact) solution is available. The techniques are of importance for numerically computed electroencephalograms (EEG's) since the numerically computed surface EEG's are severely affected by the relatively low conductivity of the compartment representing the skull.