Brain Electromagnetic Tomography Research Articles

Bayesian adaptive beamformer for robust electromagnetic brain imaging of correlated sources in high spatial resolution.

Reconstructing complex brain source activity at a high spatiotemporal resolution from magnetoencephalography (MEG) or electroencephalography (EEG) remains a challenging problem. Adaptive beamformers are routinely deployed for this imaging domain using the sample data covariance. However adaptive beamformers have long been hindered by 1) high degree of correlation between multiple brain sources, and 2) interference and noise embedded in sensor measurements. This study develops a novel framework for minimum variance adaptive beamformers that uses a model data covariance learned from data using a sparse Bayesian learning algorithm (SBL-BF). The learned model data covariance effectively removes influence from correlated brain sources and is robust to noise and interference without the need for baseline measurements. A multiresolution framework for model data covariance computation and parallelization of the beamformer implementation enables efficient high-resolution reconstruction images. Results with both simulations and real datasets indicate that multiple highly correlated sources can be accurately reconstructed, and that interference and noise can be sufficiently suppressed. Reconstructions at 2-2.5mm resolution (~150K voxels) are possible with efficient run times of 1-3 minutes. This novel adaptive beamforming algorithm significantly outperforms the state-of-the-art benchmarks. Therefore, SBL-BF provides an effective framework for efficiently reconstructing multiple correlated brain sources with high resolution and robustness to interference and noise.

Bayesian Algorithms for Joint Estimation of Brain Activity and Noise in Electromagnetic Imaging.

Simultaneously estimating brain source activity and noise has long been a challenging task in electromagnetic brain imaging using magneto- and electroencephalography. The problem is challenging not only in terms of solving the NP-hard inverse problem of reconstructing unknown brain activity across thousands of voxels from a limited number of sensors, but also for the need to simultaneously estimate the noise and interference. We present a generative model with an augmented leadfield matrix to simultaneously estimate brain source activity and sensor noise statistics in electromagnetic brain imaging (EBI). We then derive three Bayesian inference algorithms for this generative model (expectation-maximization (EBI-EM), convex bounding (EBI-Convex) and fixed-point (EBI-Mackay)) to simultaneously estimate the hyperparameters of the prior distribution for brain source activity and sensor noise. A comprehensive performance evaluation for these three algorithms is performed. Simulations consistently show that the performance of EBI-Convex and EBI-Mackay updates is superior to that of EBI-EM. In contrast to the EBI-EM algorithm, both EBI-Convex and EBI-Mackay updates are quite robust to initialization, and are computationally efficient with fast convergence in the presence of both Gaussian and real brain noise. We also demonstrate that EBI-Convex and EBI-Mackay update algorithms can reconstruct complex brain activity with only a few trials of sensor data, and for resting-state data, achieving significant improvement in source reconstruction and noise learning for electromagnetic brain imaging.

A short computerized cognitive training may affect cortical sources of rsEEG rhythms in Alzheimer’s disease patients

AbstractBackgroundPrevious evidence showed abnormal posterior sources of resting‐state electroencephalographic (rsEEG) delta (<4 Hz) and alpha (about 8‐12 Hz) rhythms in patients with Alzheimer's disease (AD) as cortical neural synchronization markers in quiet wakefulness. We tested whether daily computerized cognitive training (CCT), for a short period, with an application of serious games, may affect these mechanisms,MethodClinical and rsEEG data were collected in AD and Parkinson's disease (PD) patients with cognitive deficits and normal elderly (Nold) subjects. The rsEEG data in the patients were recorded before and after the two weeks of computerized cognitive training. Cortical rsEEG sources were estimated by exact low‐resolution brain electromagnetic tomography (eLORETA).ResultAD and PD patients were both characterized by increased performance scores in the execution of the neuropsychological tasks after the CCT program. In AD patients, no statistical difference (p >0.05) was observed in the rsEEG markers after the CCT program. In PD patients, the widespread delta source activities significantly decreased after cognitive training.ConclusionThese preliminary results suggest that the CCT program with this application of serious games differently affects the cortical neural synchronization mechanisms of rsEEG of the AD and PD patients. In the AD patients, it may have no effect, possibly because of the severely compromised integrity of the cholinergic system in those patients.

Unification of sparse Bayesian learning algorithms for electromagnetic brain imaging with the majorization minimization framework

Methods for electro- or magnetoencephalography (EEG/MEG) based brain source imaging (BSI) using sparse Bayesian learning (SBL) have been demonstrated to achieve excellent performance in situations with low numbers of distinct active sources, such as event-related designs. This paper extends the theory and practice of SBL in three important ways. First, we reformulate three existing SBL algorithms under the majorization-minimization (MM) framework. This unification perspective not only provides a useful theoretical framework for comparing different algorithms in terms of their convergence behavior, but also provides a principled recipe for constructing novel algorithms with specific properties by designing appropriate bounds of the Bayesian marginal likelihood function. Second, building on the MM principle, we propose a novel method called LowSNR-BSI that achieves favorable source reconstruction performance in low signal-to-noise-ratio (SNR) settings. Third, precise knowledge of the noise level is a crucial requirement for accurate source reconstruction. Here we present a novel principled technique to accurately learn the noise variance from the data either jointly within the source reconstruction procedure or using one of two proposed cross-validation strategies. Empirically, we could show that the monotonous convergence behavior predicted from MM theory is confirmed in numerical experiments. Using simulations, we further demonstrate the advantage of LowSNR-BSI over conventional SBL in low-SNR regimes, and the advantage of learned noise levels over estimates derived from baseline data. To demonstrate the usefulness of our novel approach, we show neurophysiologically plausible source reconstructions on averaged auditory evoked potential data.

Robust estimation of noise for electromagnetic brain imaging with the champagne algorithm

Robust estimation of the number, location, and activity of multiple correlated brain sources has long been a challenging task in electromagnetic brain imaging from M/EEG data, one that is significantly impacted by interference from spontaneous brain activity, sensor noise, and other sources of artifacts. Recently, we introduced the Champagne algorithm, a novel Bayesian inference algorithm that has shown tremendous success in M/EEG source reconstruction. Inherent to Champagne and most other related Bayesian reconstruction algorithms is the assumption that the noise covariance in sensor data can be estimated from “baseline” or “control” measurements. However, in many scenarios, such baseline data is not available, or is unreliable, and it is unclear how best to estimate the noise covariance. In this technical note, we propose several robust methods to estimate the contributions to sensors from noise arising from outside the brain without the need for additional baseline measurements. The incorporation of these methods for diagonal noise covariance estimation improves the robust reconstruction of complex brain source activity under high levels of noise and interference, while maintaining the performance features of Champagne. Specifically, we show that the resulting algorithm, Champagne with noise learning, is quite robust to initialization and is computationally efficient. In simulations, performance of the proposed noise learning algorithm is consistently superior to Champagne without noise learning. We also demonstrate that, even without the use of any baseline data, Champagne with noise learning is able to reconstruct complex brain activity with just a few trials or even a single trial, demonstrating significant improvements in source reconstruction for electromagnetic brain imaging.

Unsupervised Algorithm for Brain Anomalies Localization in Electromagnetic Imaging

A brain anomaly localization algorithm in an unsupervised machine learning (ML) framework is presented for electromagnetic brain imaging. The method is based on expected value estimation and takes the advantage of the highly symmetrical human brain. The algorithm processes signals collected from pairs of antennas that are positioned symmetrically around the head, discretizes the imaging domain into pixels, and computes the statistical fields between the antennas on the left and right sides of the head. Then, it concatenates their intensities along the axis normal to the imaging domain to compute the expected value for every pixel. The computed expected values are merged into a matrix containing expected values for all pixels. Pixels with higher intensity show the likelihood of an anomaly being present at that location. The assumption on brain symmetry from the electromagnetic perspective was tested on healthy volunteers using a 14-element array system with a working frequency band of 0.5 - 2.0 GHz. The obtained average similarity is 92% and it confirms the validity of the assumption. The same system is used to test the algorithm on different scenarios in simulations and experiments using realistic 3D head phantoms designed based on MRIs of real patients. The imaging results demonstrate the capability of the proposed algorithm to localize bleeding and estimate its size with less than 10% error in less than a minute, which makes it suitable for real-time use in emergency stroke scenarios.

Zeffiro User Interface for Electromagnetic Brain Imaging: a GPU Accelerated FEM Tool for Forward and Inverse Computations in Matlab

This article introduces the Zeffiro interface (ZI) version 2.2 for brain imaging. ZI aims to provide a simple, accessible and multimodal open source platform for finite element method (FEM) based and graphics processing unit (GPU) accelerated forward and inverse computations in the Matlab environment. It allows one to (1) generate a given multi-compartment head model, (2) to evaluate a lead field matrix as well as (3) to invert and analyze a given set of measurements. GPU acceleration is applied in each of the processing stages (1)–(3). In its current configuration, ZI includes forward solvers for electro-/magnetoencephalography (EEG) and linearized electrical impedance tomography (EIT) as well as a set of inverse solvers based on the hierarchical Bayesian model (HBM). We report the results of EEG and EIT inversion tests performed with real and synthetic data, respectively, and demonstrate numerically how the inversion parameters affect the EEG inversion outcome in HBM. The GPU acceleration was found to be essential in the generation of the FE mesh and the LF matrix in order to achieve a reasonable computing time. The code package can be extended in the future based on the directions given in this article.

Robust Empirical Bayesian Reconstruction of Distributed Sources for Electromagnetic Brain Imaging.

Electromagnetic brain imaging is the reconstruction of brain activity from non-invasive recordings of the magnetic fields and electric potentials. An enduring challenge in this imaging modality is estimating the number, location, and time course of sources, especially for the reconstruction of distributed brain sources with complex spatial extent. Here, we introduce a novel robust empirical Bayesian algorithm that enables better reconstruction of distributed brain source activity with two key ideas: kernel smoothing and hyperparameter tiling. Since the proposed algorithm builds upon many of the performance features of the sparse source reconstruction algorithm - Champagne and we refer to this algorithm as Smooth Champagne. Smooth Champagne is robust to the effects of high levels of noise, interference, and highly correlated brain source activity. Simulations demonstrate excellent performance of Smooth Champagne when compared to benchmark algorithms in accurately determining the spatial extent of distributed source activity. Smooth Champagne also accurately reconstructs real MEG and EEG data.

P68-F Abnormalities of cortical neural synchronization mechanisms in patients with Alzheimer’s diseases dementia: An EEG study

Background This retrospective study aimed to check whether the resting state eyes-closed electroencephalographic (rsEEG) rhythms might reflect brain arousal in patients with Alzheimer’s disease dementia (ADD) at the group and the individual level. Methods Our dataset was an international archive of Clinical and rsEEG data of 42 ADD and 40 healthy elderly (Nold) subjects. The delta, theta, alpha 1, alpha 2, and alpha 3 frequency bands were determined by Individual alpha frequency peak (IAF). Fixed beta 1, beta 2, and gamma bands were also considered. rsEEG cortical sources were estimated by means of the exact low-resolution brain electromagnetic source tomography and were then classified across individuals, on the basis of the receiver operating characteristic (ROC) curves. Results Compared to the Nold group, IAF showed slowing in the ADD group. Furthermore, The ADD group showed lower posterior alpha 2 source activities and higher occipital delta source activities. The ratio of occipital delta to alpha 2 allowed good classification accuracy (sensitivity: 88.1%, specificity: 85%, and accuracy: 86.6%) between the Nold and ADD individuals. Conclusion In quiet wakefulness, delta and alpha sources unveiled different spatial and frequency features of the cortical neural synchronization underpinning brain arousal in ADD patients compared with Nold subjects. An ongoing study is testing the enrichment of markers derived from structural magnetic resonance imaging (MRI) in the classification accuracy between Nold and ADD patients for potential clinical applications.

Source localization of epileptiform discharges in childhood absence epilepsy using a distributed source model: a standardized, low-resolution, brain electromagnetic tomography (sLORETA) study.

Localizing the source of epileptiform discharges in generalized epilepsy has been controversial for the past few decades. Recent neuroimaging studies have shown that epileptiform discharges in generalized epilepsy can be localized to a particular region. Childhood absence epilepsy (CAE) is the most common generalized epilepsy in childhood and is considered the prototype of idiopathic generalized epilepsy (IGE). To better understand electrophysiological changes and their development in CAE, we investigated the origin of epileptiform discharges. We performed distributed source localization with standardized, low-resolution, brain electromagnetic tomography (sLORETA). In 16 children with CAE, sLORETA images corresponding to the midpoint of the ascending phase and the negative peak of the spike were obtained from a total of 242 EEG epochs (121 epochs at each timepoint). Maximal current source density (CSD) was mostly located in the frontal lobe (69.4%). At the gyral level, maximal CSD was most commonly in the superior frontal gyrus (39.3%) followed by the middle frontal gyrus (14.0%) and medial frontal gyrus (8.7%). At the hemisphere level, maximal CSD was dominant in the right cerebral hemisphere (63.6%). These results were consistent at the midpoint of the ascending phase and the negative peak of the spike. Our results demonstrated that the major source of epileptiform discharges in CAE was the frontal lobe. These results suggest that the frontal lobe is involved in generating CAE. This finding is consistent with recent studies that have suggested selective cortical involvement, especially in the frontal regions, in IGE.

Hierarchical multiscale Bayesian algorithm for robust MEG/EEG source reconstruction

In this paper, we present a novel hierarchical multiscale Bayesian algorithm for electromagnetic brain imaging using magnetoencephalography (MEG) and electroencephalography (EEG). In particular, we present a solution to the source reconstruction problem for sources that vary in spatial extent. We define sensor data measurements using a generative probabilistic graphical model that is hierarchical across spatial scales of brain regions and voxels. We then derive a novel Bayesian algorithm for probabilistic inference with this graphical model. This algorithm enables robust reconstruction of sources that have different spatial extent, from spatially contiguous clusters of dipoles to isolated dipolar sources. We compare the new algorithm with several representative benchmarks on both simulated and real brain activities. The source locations and the correct estimation of source time courses used for the simulated data are chosen to test the performance on challenging source configurations. In simulations, performance of the novel algorithm shows superiority to several existing benchmark algorithms. We also demonstrate that the new algorithm is more robust to correlated brain activity present in real MEG and EEG data and is able to resolve distinct and functionally relevant brain areas with real MEG and EEG datasets.

EEG‐based brain source localization using visual stimuli

ABSTRACTElectroencephalography (EEG) is widely used in variety of research and clinical applications which includes the localization of active brain sources. Brain source localization provides useful information to understand the brain's behavior and cognitive analysis. Various source localization algorithms have been developed to determine the exact locations of the active brain sources due to which electromagnetic activity is generated in brain. These algorithms are based on digital filtering, 3D imaging, array signal processing and Bayesian approaches. According to the spatial resolution provided, the algorithms are categorized as either low resolution methods or high resolution methods. In this research study, EEG data is collected by providing visual stimulus to healthy subjects. FDM is used for head modelling to solve forward problem. The low‐resolution brain electromagnetic tomography (LORETA) and standardized LORETA (sLORETA) have been used as inverse modelling methods to localize the active regions in the brain during the stimulus provided. The results are produced in the form of MRI images. The tables are also provided to describe the intensity levels for estimated current level for the inverse methods used. The higher current value or intensity level shows the higher electromagnetic activity for a particular source at certain time instant. Thus, the results obtained demonstrate that standardized method which is based on second order Laplacian (sLORETA) in conjunction with finite difference method (FDM) as head modelling technique outperforms other methods in terms of source estimation as it has higher current level and thus, current density (J) for an area as compared to others.

Distinct β Band Oscillatory Networks Subserving Motor and Cognitive Control during Gait Adaptation

Everyday locomotion and obstacle avoidance requires effective gait adaptation in response to sensory cues. Many studies have shown that efficient motor actions are associated with μ rhythm (8-13 Hz) and β band (13-35 Hz) local field desynchronizations in sensorimotor and parietal cortex, whereas a number of cognitive task studies have reported higher behavioral accuracy to be associated with increases in β band power in prefrontal and sensory cortex. How these two distinct patterns of β band oscillations interplay during gait adaptation, however, has not been established. Here we recorded 108 channel EEG activity from 18 participants (10 males, 22-35 years old) attempting to walk on a treadmill in synchrony with a series of pacing cue tones, and quickly adapting their step rate and length to sudden shifts in pacing cue tempo. Independent component analysis parsed each participant's EEG data into maximally independent component (IC) source processes, which were then grouped across participants into distinct spatial/spectral clusters. Following cue tempo shifts, mean β band power was suppressed for IC sources in central midline and parietal regions, whereas mean β band power increased in IC sources in or near medial prefrontal and dorsolateral prefrontal cortex. In the right dorsolateral prefrontal cortex IC cluster, the β band power increase was stronger during (more effortful) step shortening than during step lengthening. These results thus show that two distinct patterns of β band activity modulation accompany gait adaptations: one likely serving movement initiation and execution; and the other, motor control and inhibition. Understanding brain dynamics supporting gait adaptation is crucial for understanding motor deficits in walking, such as those associated with aging, stroke, and Parkinson's. Only a few electromagnetic brain imaging studies have examined neural correlates of human upright walking. Here, application of independent component analysis to EEG data recorded during treadmill walking allowed us to uncover two distinct β band oscillatory cortical networks that are active during gait adaptation to shifts in the tempo of an auditory pacing cue: (8-13 Hz) μ rhythm and (13-35 Hz) β band power decreases in central and parietal cortex and (14-20 Hz) β band power increases in frontal brain areas. These results provide a fuller framework for electrophysiological studies of cortical gait control and its disorders.

Detection of activity from the amygdala with magnetoencephalography

Combining excellent temporal resolution with accurate source localization is a challenge for brain imaging studies in humans. In this domain, showing evidence for the contribution of deep brain structures to magneto-encephalographic (MEG) signal may represent an important step. The amygdala, is an example of such structure that is essential for social cognition and emotional processes. To date, MEG studies have been mainly focused on cortical activities. Under the impulsion of Line Garnero, we tackled the challenge of detecting the activation of deep brain structures with non-invasive electromagnetic brain imaging methods and developed a new method to localize amygdala sources. Here, we first review amygdala structural and histological properties that argue in favour of the detection of its activity with MEG. Then we present a simulation study of amygdala contribution to MEG signal. This study was based on an original anatomical segmentation technique, which allowed distributing sources within the amygdala volume of each subject. Finally, we present preliminary data from a face perception experiment where amygdala activity was examined with this new distributed source imaging method.

Quantitative EEG and Electromagnetic Brain Imaging in Aging and in the Evolution of Dementia

Electroencephalographic (EEG) changes with normal aging have long been reported. Departures from age-expected changes have been observed in mild cognitive impairment and dementia, the magnitude of which correlates with the degree of cognitive impairment. Such abnormalities include increased delta and theta activity, decreased mean frequency, and changes in coherence. Similar findings have been reported using magnetoencephalography (MEG) at rest and during performance of mental tasks. Electrophysiological features have also been shown to be predictive of future decline in mild cognitive impairment (MCI) and Alzheimer's disease (AD). We have recently reported results from initial quantitative electroencephalography (QEEG) evaluations of normal elderly subjects (with only subjective reports of memory loss), predicting future cognitive decline or conversion to dementia, with high prediction accuracy (approximately 95%). In this report, source localization algorithms were used to identify the mathematically most probable underlying generators of abnormal features of the scalp-recorded EEG from these patients with differential outcomes. Using this QEEG method, abnormalities in brain regions identified in studies of AD using MEG, MRI, and positron emission tomography (PET) imaging were found in the premorbid recordings of those subjects who go on to decline or convert to dementia.

Linear Inverse Solutions: Simulations from a Realistic Head Model in MEG

Distributed linear solutions are widely used in source localization to solve the ill-posed EEG/MEG inverse problem. In the classical approach based on dipole sources, these methods estimate the current densities at a great number of brain sites, typically at the nodes of a 3-D grid which discretizes the chosen solution space. The estimated current density distributions are displayed as brain electromagnetic tomography (BET) images. We have tested well known minimum norm solutions (MN, WMN, LORETA) and other linear inverse solutions [WROP, sLORETA, interference uniform, gain uniform, weight vector normalized (WVN), and a new solution named SLF (Standardized Lead Field)], using a MEG configuration (BTi Magnes 2500 WH with 148 axial magnetometers) and a realistic head model using BEM (Boundary Element Method). The solutions were compared in a noise-free condition and in the presence of noise using the classical dipole localization errors (DLE) together with a new figure of merit that we called max gain uniformity, which measures the capability of an inverse linear solution to show spots of activity with similar amplitudes on the brain electromagnetic tomographies when multiple dipole sources with similar moments are simultaneously active. Whereas some solutions (sLORETA, interference uniform and SLF) were capable of zero dipole localization errors in the noise-free case, none of them reached 100% of correct dipole localizations in the presence of a high level of Gaussian noise. The SLF solution, which has the advantage to be independent from any regularization parameter, presented the best results with the lowest max gain uniformities, with almost 100% of correct dipole localizations with 10% of noise and more than 90% of correct localizations with 30% of noise added to the data. Nevertheless, no solution was able to combine at the same time a correct localization of single sources and the capability to visualize multiple sources with comparable amplitudes on the brain electromagnetic tomographies.

Electrical Impedance Tomography

t. This paper surveys some of the work our group has done in electrical impedance tomography.