Multi-Resolution FOCUSS: A Source Imaging Technique Applied to MEG Data

Summary: A variety of techniques are available for imaging magnetoencephalographic (MEG) data to the corresponding cortical structures. Each performs a functional optimization that includes mathematical and physical restrictions on source activity. Unlike other imaging techniques, MR-FOCUSS (Multi-Resolution FOCal Underdetermined System Solution) utilizes a wavelet statistical operator that allows spatial resolution to be chosen appropriately for focal or extended sources. Control of focal imaging properties is achieved by specifying P in an l_P norm distribution template used to construct the wavelets. In addition, incorporation of a multi-resolution wavelet operator desensitizes the mathematical algorithm to noise, (regularization). Like the FOCUSS imaging technique, an initial estimate of cortical activity is recursively enhanced to obtain the final high resolution imaging results. Studies of model MEG data representing all regions of a realistic cortical model are performed to quantify MR-FOCUSS imaging properties. These modeled data studies included single and multiple dipole sources as well as an extended source model. Thus, MR-FOCUSS is found to be very effective for imaging language processing for pre-surgical planning and provides a high-resolution method to image sequential activation of multiple correlated sources involved in language processing. This research Including the development of the MR-FOCUSS imaging technique and software implementation was supported by NIH/NINDS Grant R01 NS30914.     

A technique to consider mismatches between fMRI and EEG/MEG sources for fMRI-constrained EEG/MEG source imaging: a preliminary simulation study

fMRI-constrained EEG/MEG source imaging can be a powerful tool in studying human brain functions with enhanced spatial and temporal resolutions. Recent studies on the combination of fMRI and EEG/MEG have suggested that fMRI prior information could be readily implemented by simply imposing different weighting factors to cortical sources overlapping with the fMRI activations. It has been also reported, however, that such a hard constraint may cause severe distortions or elimination of meaningful EEG/MEG sources when there are distinct mismatches between the fMRI activations and the EEG/MEG sources. If one wants to obtain the actual EEG/MEG source locations and uses the fMRI prior information as just an auxiliary tool to enhance focality of the distributed EEG/MEG sources, it is reasonable to weaken the strength of fMRI constraint when severe mismatches between fMRI and EEG/MEG sources are observed. The present study suggests an efficient technique to automatically adjust the strength of fMRI constraint according to the mismatch level. The use of the proposed technique rarely affects the results of conventional fMRI-constrained EEG/MEG source imaging if no major mismatch between the two modalities is detected; while the new results become similar to those of typical EEG/MEG source imaging without fMRI constraint if the mismatch level is significant. A preliminary simulation study using realistic EEG signals demonstrated that the proposed technique can be a promising tool to selectively apply fMRI prior information to EEG/MEG source imaging.     

Anatomically constrained minimum variance beamforming applied to EEG

Neural activity as measured non-invasively using electroencephalography (EEG) or magnetoencephalography (MEG) originates in the cortical gray matter. In the cortex, pyramidal cells are organized in columns and activated coherently, leading to current flow perpendicular to the cortical surface. In recent years, beamforming algorithms have been developed, which use this property as an anatomical constraint for the locations and directions of potential sources in MEG data analysis. Here, we extend this work to EEG recordings, which require a more sophisticated forward model due to the blurring of the electric current at tissue boundaries where the conductivity changes. Using CT scans, we create a realistic three-layer head model consisting of tessellated surfaces that represent the cerebrospinal fluid-skull, skull-scalp, and scalp-air boundaries. The cortical gray matter surface, the anatomical constraint for the source dipoles, is extracted from MRI scans. EEG beamforming is implemented on simulated sets of EEG data for three different head models: single spherical, multi-shell spherical, and multi-shell realistic. Using the same conditions for simulated EEG and MEG data, it is shown (and quantified by receiver operating characteristic analysis) that EEG beamforming detects radially oriented sources, to which MEG lacks sensitivity. By merging several techniques, such as linearly constrained minimum variance beamforming, realistic geometry forward solutions, and cortical constraints, we demonstrate it is possible to localize and estimate the dynamics of dipolar and spatially extended (distributed) sources of neural activity.     

MEG and EEG Sensitivity in a Case of Medial Occipital Epilepsy

Interictal or ictal events in partial epilepsies may project on scalp EEG contralaterally to the side of the epileptogenic lesion. Such paradoxical lateralization can be observed in case of para-sagittal generators, and is likely due to the spatial orientation of the generator, presenting an oblique projection towards the midline. We present here a case of medial occipital epilepsy investigated using EEG, MEG and stereoelectroencephalography (SEEG). MRI displayed a focal cortical dysplasia in the superior margin of the right calcarine fissure. SEEG demonstrated bilateral medial occipital interictal spikes, with an inversion of polarity at the level of the lesion and a contralateral propagation occurring in 10 ms. Interictal iterative EEG cartographies showed a large posterior field, with a maximum contralateral to the initial generator (EEG paradoxical lateralization). With the same number of channels, interictal iterative MEG cartographies were more precise and more complex than EEG ones, indicating an onset accurately lateralized. A few milliseconds later, MEG cartographies were quadripolar, thus indicating two homotopic active generators. These MEG and EEG cartographies have been reproduced using BESA dipole simulator. Relative merits of MEG and EEG are still debated. With 151 channels, MEG source localizations indicated the right medial occipital area, as demonstrated by SEEG. An investigation with a corresponding number of EEG channels was not performed. After a down sampling to 64 sensors, this precision was lost. MEG and EEG source localization results, both with 64 channels, were quite comparable, indicating both medial occipital areas. However, a careful analysis of MEG/EEG iterative cartographies, performed with the same number of channels in both modalities, demonstrated that, in this configuration, MEG sensitivity was superior to the EEG one, allowing separating two medial occipital sources, characterized in SEEG by a time delay of 10 ms.     

Evaluation of smoothing in an iterative lp-norm minimization algorithm for surface-based source localization of MEG

The imaging of neural sources of magnetoencephalographic data based on distributed source models requires additional constraints on the source distribution in order to overcome ill-posedness and obtain a plausible solution. The minimum l(p) norm (0 < p < or = 1) constraint is known to be appropriate for reconstructing focal sources distributed in several regions. A well-known recursive method for solving the l(p)-norm minimization problem, for example, is the focal underdetermined system solver (FOCUSS). However, this iterative algorithm tends to give spurious sources when the noise level is high. In this study, we present an algorithm to incorporate a smoothing technique into the FOCUSS algorithm and test different smoothing kernels in a surface-based cortical source space. Simulations with cortical source patches assumed in auditory areas show that the incorporation of the smoothing procedure improves the performance of the FOCUSS algorithm, and that using the geodesic distance for constructing a smoothing kernel is a better choice than using the Euclidean one, particularly when employing a cortical source space. We also apply these methods to a real data set obtained from an auditory experiment and illustrate their applicability to realistic data by presenting the reconstructed source images localized in the superior temporal gyrus.     

MRI prior computation and parallel tempering algorithm: a probabilistic resolution of the MEG/EEG inverse problem

Since the MEG inverse problem is ill-posed and admits many possible solutions, it is not possible to give it a single " true" answer. Therefore, we propose here to use a specific probabilistic algorithm to map the full probability distribution of the MEG sources with Markov Chain Monte Carlo methods. Using a Bayesian approach, the probability of the MEG solutions is expressed as the product of the likelihood by the prior probability. To compute the prior and constrain the MEG inverse problem resolution, MRI data are also acquired and automatically processed to determine the brain position and volume. We then use Parallel Tempering algorithm to estimate the full posterior probability and determine the likely solutions of the inverse problem. We illustrate the method with results obtained from the analysis of somatosensory data. This illustrates both the MRI processing for the prior computation, and how the knowledge of the full posterior probability distribution can be used to estimate the position of the sources, as well as their likely extension.     

Zoomed MRI Guided by Combined EEG/MEG Source Analysis: A Multimodal Approach for Optimizing Presurgical Epilepsy Work-up and its Application in a Multi-focal Epilepsy Patient Case Study

In recent years, the use of source analysis based on electroencephalography (EEG) and magnetoencephalography (MEG) has gained considerable attention in presurgical epilepsy diagnosis. However, in many cases the source analysis alone is not used to tailor surgery unless the findings are confirmed by lesions, such as, e.g., cortical malformations in MRI. For many patients, the histology of tissue resected from MRI negative epilepsy shows small lesions, which indicates the need for more sensitive MR sequences. In this paper, we describe a technique to maximize the synergy between combined EEG/MEG (EMEG) source analysis and high resolution MRI. The procedure has three main steps: (1) construction of a detailed and calibrated finite element head model that considers the variation of individual skull conductivities and white matter anisotropy, (2) EMEG source analysis performed on averaged interictal epileptic discharges (IED), (3) high resolution (0.5 mm) zoomed MR imaging, limited to small areas centered at the EMEG source locations. The proposed new diagnosis procedure was then applied in a particularly challenging case of an epilepsy patient: EMEG analysis at the peak of the IED coincided with a right frontal focal cortical dysplasia (FCD), which had been detected at standard 1 mm resolution MRI. Of higher interest, zoomed MR imaging (applying parallel transmission, ‘ZOOMit’) guided by EMEG at the spike onset revealed a second, fairly subtle, FCD in the left fronto-central region. The evaluation revealed that this second FCD, which had not been detectable with standard 1 mm resolution, was the trigger of the seizures.     

Anatomically constrained dipole adjustment (ANACONDA) for accurate MEG/EEG focal source localizations

This paper proposes an alternative approach to enhance localization accuracy of MEG and EEG focal sources. The proposed approach assumes anatomically constrained spatio-temporal dipoles, initial positions of which are estimated from local peak positions of distributed sources obtained from a pre-execution of distributed source reconstruction. The positions of the dipoles are then adjusted on the cortical surface using a novel updating scheme named cortical surface scanning. The proposed approach has many advantages over the conventional ones: (1) as the cortical surface scanning algorithm uses spatio-temporal dipoles, it is robust with respect to noise; (2) it requires no a priori information on the numbers and initial locations of the activations; (3) as the locations of dipoles are restricted only on a tessellated cortical surface, it is physiologically more plausible than the conventional ECD model. To verify the proposed approach, it was applied to several realistic MEG/EEG simulations and practical experiments. From the several case studies, it is concluded that the anatomically constrained dipole adjustment (ANACONDA) approach will be a very promising technique to enhance accuracy of focal source localization which is essential in many clinical and neurological applications of MEG and EEG.     

Size matters: MEG empirical and simulation study on source localization of the earliest visual activity in the occipital cortex

While the relationship between sensory stimulation and tasks and the size of the cortical activations is generally unknown, the visual modality offers a unique possibility of an experimental manipulation of stimulus size-related increases of the spatial extent of cortical activation even during the earliest activity in the retinotopically organized primary visual cortex. We used magnetoecephalography (MEG), visual stimuli of increasing size, and numerical simulations on realistic cortical surfaces to explore the effects of increasing spatial extent of the activated cortical sources on the neuromagnetic fields, location estimation biases, and source resolution. Source localization was performed assuming multiple dipoles in a sphere model using an efficient, automatically restarted multi-start simplex minimizer within the Calibrated Start Spatio-Temporal (CSST) algorithm. We found size-related effects on amplitude and latencies and differences in relative locations of the earliest occipital sources evoked by stimuli of increasing size presented at the same eccentricity. This finding was confirmed by single patch simulations. Additionally, simulations of multiple extended sources demonstrated size-related increase in limits in source resolution for bilaterally simulated sources, biases in location estimates for a given separation of sources, and limits in source resolution due to source multiplicity within a hemisphere.     

Magnetoencephalography-guided epilepsy surgery for children with intractable focal epilepsy: SickKids experience.

We introduced magnetoencephalography (MEG)-guided epilepsy surgery for children with intractable focal epilepsy at The Hospital for Sick Children (SickKids) in Toronto, Canada. Surgical candidacy and decisions on surgical procedure for children with intractable focal epilepsy are based on long-term scalp video EEG (VEEG) results, magnetic resonance imaging (MRI) findings, and the distribution of MEG spike sources. After multidisciplinary discussion at the seizure conference, for the patients requiring intracranial VEEG, custom-made subdural electrode grids are designed using three-dimensional MRI superimposed with MEG spike sources to cover the area of clustered MEG spike sources. At the first surgery, neurosurgeons use the intraoperative neuronavigation system to visualize the area of clustered spike dipoles and somatosensory evoked fields on MEG to place the subdural grid and depth electrodes. At the second surgery, the area of seizure onset and active interictal spike discharges on the intracranial VEEG recording, which usually correlates with the zone of clustered MEG spike sources, is resected. This combination leads to successful surgical outcome to control seizures in these challenging paediatric patients. MEG is a useful tool in children with intractable focal epilepsy to determine the surgical candidacy and focal cortical resection to stop seizures.     

Moving Mesh Method for Reconstructing Some Spread Sources in the Brain

The purpose of this paper is to propose a new algorithm for the analysis of biomagnetic field data obtained from magnetoencephalography (MEG) measurements. This new method overcomes two major problems faced by the current method of data analysis. The first problem is the need to determine the number of sites of brain activity before calculations can be performed. The second problem is inability of the analysis to provide any information regarding the volume of the brain activity. The new data analysis method, called the Moving Mesh Method (MMM), is capable of analyzing MEG data without the need to determine the number of sources beforehand. In addition, the MMM determines the location of brain activity as a three dimensional volume, instead of as a point source of activity. The MMM uses an iterative method of calculating the position of the sources to achieve greater accuracy, and a regularized g-inverse matrix to stabilize its solution. The feasibility of the MMM was examined by two methods. First, a computer simulation was used to confirm the MMM's capability to analyzing MEG data. In the second experiment, the MMM was applied to analyze somatosensory evoked fields obtained using a new imaging system (Shimadzu Biomagnetic Imaging System, Model-100). From the interpretation of the results, we have concluded that the MMM is a feasible method of biomagnetic data analysis.     

The Artifact-Free MEG-MUSIC Algorithm on Magnetoencephalographic Source Localization

INTRODUCTION　　 Multichannel superconducting quantum interference device ( SQUID) magne-tometers can beused to measure the spatio-temporal magnetoencephalogram ( MEG)produced by the neural activity in the human brain.From analysis of the MEG,onecan obtai…     

Localization Accuracy of Distributed Inverse Solutions for Electric and Magnetic Source Imaging of Interictal Epileptic Discharges in Patients with Focal Epilepsy

Distributed inverse solutions aim to realistically reconstruct the origin of interictal epileptic discharges (IEDs) from noninvasively recorded electroencephalography (EEG) and magnetoencephalography (MEG) signals. Our aim was to compare the performance of different distributed inverse solutions in localizing IEDs: coherent maximum entropy on the mean (cMEM), hierarchical Bayesian implementations of independent identically distributed sources (IID, minimum norm prior) and spatially coherent sources (COH, spatial smoothness prior). Source maxima (i.e., the vertex with the maximum source amplitude) of IEDs in 14 EEG and 19 MEG studies from 15 patients with focal epilepsy were analyzed. We visually compared their concordance with intracranial EEG (iEEG) based on 17 cortical regions of interest and their spatial dispersion around source maxima. Magnetic source imaging (MSI) maxima from cMEM were most often confirmed by iEEG (cMEM: 14/19, COH: 9/19, IID: 8/19 studies). COH electric source imaging (ESI) maxima co-localized best with iEEG (cMEM: 8/14, COH: 11/14, IID: 10/14 studies). In addition, cMEM was less spatially spread than COH and IID for ESI and MSI (p < 0.001 Bonferroni-corrected post hoc t test). Highest positive predictive values for cortical regions with IEDs in iEEG could be obtained with cMEM for MSI and with COH for ESI. Additional realistic EEG/MEG simulations confirmed our findings. Accurate spatially extended sources, as found in cMEM (ESI and MSI) and COH (ESI) are desirable for source imaging of IEDs because this might influence surgical decision. Our simulations suggest that COH and IID overestimate the spatial extent of the generators compared to cMEM.     

Evaluation of Algorithms for Intracranial EEG (iEEG) Source Imaging of Extended Sources: Feasibility of Using iEEG Source Imaging for Localizing Epileptogenic Zones in Secondary Generalized Epilepsy

Precise identification of epileptogenic zones in patients with intractable drug-resistant epilepsy is critical for successful epilepsy surgery. Numerous source-imaging algorithms for localizing epileptogenic zones based on scalp electroencephalography (EEG) and magnetoencephalography (MEG) have been developed and validated in simulation and experimental studies. Recently, intracranial EEG (iEEG)-based imaging of epileptogenic sources has attracted interest as a promising tool for presurgical evaluation of epilepsy; however, most iEEG studies have focused on localization of epileptogenic zones in focal epilepsy. In the present study, we investigated whether iEEG source imaging is a useful supplementary tool for identifying extended epileptogenic sources in secondary generalized epilepsy such as Lennox-Gastaut syndrome (LGS). To this end, we applied four different cortical source imaging algorithms, namely minimum norm estimation (MNE), low-resolution electromagnetic tomography (LORETA), standardized LORETA (sLORETA), and L _p -norm estimation (p = 1.5, referred to as Lp1.5), to artificial iEEG datasets generated assuming various source sizes and locations. We also applied these four algorithms to clinical ictal iEEG recordings acquired from a pediatric patient with LGS. Interestingly, the traditional MNE algorithm outperformed the other imaging algorithms in most of our experiments, particularly in cases when larger-sized sources were activated. Although sLORETA outperformed both LORETA and Lp1.5, its performance was not as good as that of MNE. Compared to the other algorithms, the performance of Lp1.5 decayed most rapidly as the source size increased. Our findings suggest that iEEG source imaging using MNE is a promising auxiliary tool for the identification of epileptogenic zones in secondary generalized epilepsy. We anticipate that our results will provide useful guidelines for selection of an appropriate imaging algorithm for iEEG source imaging studies.     

Sensitivity of MEG and EEG to Source Orientation

An important difference between magnetoencephalography (MEG) and electroencephalography (EEG) is that MEG is insensitive to radially oriented sources. We quantified computationally the dependency of MEG and EEG on the source orientation using a forward model with realistic tissue boundaries. Similar to the simpler case of a spherical head model, in which MEG cannot see radial sources at all, for most cortical locations there was a source orientation to which MEG was insensitive. The median value for the ratio of the signal magnitude for the source orientation of the lowest and the highest sensitivity was 0.06 for MEG and 0.63 for EEG. The difference in the sensitivity to the source orientation is expected to contribute to systematic differences in the signal-to-noise ratio between MEG and EEG.     

Face activated neurodynamic cortical networks

Previous neuroimaging studies have shown that complex visual stimuli, such as faces, activate multiple brain regions, yet little is known on the dynamics and complexity of the activated cortical networks during the entire measurable evoked response. In this study, we used simulated and face-evoked empirical MEG data from an oddball study to investigate the feasibility of accurate, efficient, and reliable spatio-temporal tracking of cortical pathways over prolonged time intervals. We applied a data-driven, semiautomated approach to spatio-temporal source localization with no prior assumptions on active cortical regions to explore non-invasively face-processing dynamics and their modulation by task. Simulations demonstrated that the use of multi-start downhill simplex and data-driven selections of time intervals submitted to the Calibrated Start Spatio-Temporal (CSST) algorithm resulted in improved accuracy of the source localization and the estimation of the onset of their activity. Locations and dynamics of the identified sources indicated a distributed cortical network involved in face processing whose complexity was task dependent. This MEG study provided the first non-invasive demonstration, agreeing with intracranial recordings, of an early onset of the activity in the fusiform face gyrus (FFG), and that frontal activation preceded parietal for responses elicited by target faces.     

Comparing Regularized and Non-Regularized Nonlinear Dipole Fit Methods: A Study in a Simulated Sulcus Structure

The inverse problem arising from EEG and MEG is largely underdetermined. One strategy to alleviate this problem is the restriction to a limited number of point-like sources, the focal source model. Although the singular value decomposition of the spatio-temporal data gives an estimate of the minimal number of dipoles contributing to the measurement, the exact number is unknown in advance and noise complicates the reconstruction. Classical non-regularized nonlinear dipole fit algorithms do not give an estimate for the correct number because they are not stable with regard to an overestimation of this parameter. Too many sources may only describe noise but can still attain a large magnitude during the inverse procedure and may be indiscernible from the true sources. This paper describes a nonlinear dipole fit reconstruction algorithm with a new regularization approach for the embedded linear problem, automatically controlled by the noise in the data and the condition of the occuring least square problems. The algorithm is stable with regard to source components which nearly lie in the kernel of the projection or lead field operator and it thus gives an estimate of the unknown number parameter. EEG simulation studies in a simulated sulcus structure are carried out for an instantaneous dipole model and spatial resolution in the sulcus and stability of the new method are compared with a classical reconstruction algorithm without regularization.     

Source Connectivity Analysis from MEG and its Application to Epilepsy Source Localization

We report an approach to perform source connectivity analysis from MEG, and initially evaluate this approach to interictal MEG to localize epileptogenic foci and analyze interictal discharge propagations in patients with medically intractable epilepsy. Cortical activities were reconstructed from MEG using individual realistic geometry boundary element method head models. Directional connectivity among cortical regions of interest was then estimated using directed transfer function. The MEG source connectivity analysis method was implemented in the eConnectome software, which is open-source and freely available at . As an initial evaluation, the method was applied to study MEG interictal spikes from five epilepsy patients. Estimated primary epileptiform sources were consistent with surgically resected regions, suggesting the feasibility of using cortical source connectivity analysis from interictal MEG for potential localization of epileptiform activities.     

CT and MRI derived source localization error in a custom prostate phantom using automated image coregistration.

Dosimetric evaluation of completed brachytherapy implant procedures is crucial in developing proper technique. Additionally, accurate dosimetry may be useful in predicting the success of an implant. Accurate definition of the prostate gland and localization of the implanted radioactive sources are critical to attain meaningful dosimetric data. MRI is recognized as a superior imaging modality in delineating the prostate gland. More importantly, MRI can be used for source localization in postimplant prostates. However, the MRI derived source localization error bears further investigation. We present a useful tool in determining the source localization error as well as permitting the fusion, or coregistration, of selected data from multiple imaging modalities. We constructed a custom prostate phantom of hydrocolloid material precisely implanted with I-125 seeds. We obtained CT, the accepted modality, and MRI scans of the phantom. Subsequently, we developed an automated algorithm that employs a sequential translation of data sets to initially maximize coregistration and minimize error between data sets. This was followed by a noniterative solution for the necessary rotation transformation matrix using the Orthogonal Procrustes Solution. We applied this algorithm to CT and MRI scans of the custom phantom. CT derived source locations had source localization errors of 1.59 mm +/- 0.64. MRI derived source locations produced similar results (1.67 mm +/- 0.76). These errors may be attributed to the image digitization process.