Shape-constrained deformable brain segmentation: Methods and quantitative validation.

MRI-guided neuro interventions require rapid, accurate, and reproducible segmentation of anatomical brain structures for identification of targets during surgical procedures and post-surgical evaluation of intervention efficiency. Segmentation algorithms must be validated and cleared for clinical use. This work introduces a methodology for shape-constrained deformable brain segmentation, describes the quantitative validation used for its clinical clearance, and presents a comparison with manual expert segmentation and FreeSurfer, an open source software for neuroimaging data analysis. ClearPoint Maestro is software for fully-automatic brain segmentation from T1-weighted MRI that combines a shape-constrained deformable brain model with voxel-wise tissue segmentation within the cerebral hemispheres and the cerebellum. The performance of the segmentation was validated in terms of accuracy and reproducibility. Segmentation accuracy was evaluated with respect to training data and independently traced ground truth. Segmentation reproducibility was quantified and compared with manual expert segmentation and FreeSurfer. Quantitative reproducibility analysis indicates superior performance compared to both manual expert segmentation and FreeSurfer. The shape-constrained methodology results in accurate and highly reproducible segmentation. Inherent point based-correspondence provides consistent target identification ideal for MRI-guided neuro interventions.

Patient-Specific Sensor Registration for Electrical Source Imaging Using a Deformable Head Model.

OBJECTIVE: Electrical source imaging of brain activity is most accurate when using individualized bioelectric head models. Constructing these models requires identifying electrode positions on the scalp surface. Current methods such as photogrammetry involve significant user interaction that limits integration in clinical workflows. This work introduces and validates a new, fully-automatic method for sensor registration. METHODS: Average electrode coordinates are registered to the mean scalp mesh of a shape-constrained deformable head model used for tissue segmentation. Patient-specific electrode positions can be identified on the deformed scalp surface using point-based correspondence after model adaptation. RESULTS: The performance of the proposed method for sensor registration is evaluated with simulated and real data. Electrode variability is quantified for a photogrammetry-based solution and compared against the proposed sensor registration. CONCLUSION: A fully-automated model-based approach can identify electrode locations with similar accuracy as a current state-of-the-art photogrammetry system. SIGNIFICANCE: The new method for sensor registration presented in this work is rapid and fully automatic. It eliminates any user dependent inaccuracy introduced in sensor registration and ensures reproducible results. More importantly, it can more easily be integrated in clinical workflows, enabling broader adoption of electrical source imaging technologies.

A Simulation Framework for Benchmarking EEG-Based Brain Connectivity Estimation Methodologies.

Due to its high temporal resolution, electroencephalography (EEG) is widely used to study functional and effective brain connectivity. Yet, there is currently a mismatch between the vastness of studies conducted and the degree to which the employed analyses are theoretically understood and empirically validated. We here provide a simulation framework that enables researchers to test their analysis pipelines on realistic pseudo-EEG data. We construct a minimal example of brain interaction, which we propose as a benchmark for assessing a methodology's general eligibility for EEG-based connectivity estimation. We envision that this benchmark be extended in a collaborative effort to validate methods in more complex scenarios. Quantitative metrics are defined to assess a method's performance in terms of source localization, connectivity detection and directionality estimation. All data and code needed for generating pseudo-EEG data, conducting source reconstruction and connectivity estimation using baseline methods from the literature, evaluating performance metrics, as well as plotting results, are made publicly available. While this article covers only EEG modeling, we will also provide a magnetoencephalography version of our framework online.

Self-Consistent MUSIC: An approach to the localization of true brain interactions from EEG/MEG data.

MUltiple SIgnal Classification (MUSIC) is a standard localization method which is based on the idea of dividing the vector space of the data into two subspaces: signal subspace and noise subspace. The brain, divided into several grid points, is scanned entirely and the grid point with the maximum consistency with the signal subspace is considered as the source location. In one of the MUSIC variants called Recursively Applied and Projected MUSIC (RAP-MUSIC), multiple iterations are proposed in order to decrease the location estimation uncertainties introduced by subspace estimation errors. In this paper, we suggest a new method called Self-Consistent MUSIC (SC-MUSIC) which extends RAP-MUSIC to a self-consistent algorithm. This method, SC-MUSIC, is based on the idea that the presence of several sources has a bias on the localization of each source. This bias can be reduced by projecting out all other sources mutually rather than iteratively. While the new method is applicable in all situations when MUSIC is applicable we will study here the localization of interacting sources using the imaginary part of the cross-spectrum due to the robustness of this measure to the artifacts of volume conduction. For an odd number of sources this matrix is rank deficient similar to covariance matrices of fully correlated sources. In such cases MUSIC and RAP-MUSIC fail completely while the new method accurately localizes all sources. We present results of the method using simulations of odd and even number of interacting sources in the presence of different noise levels. We compare the method with three other source localization methods: RAP-MUSIC, dipole fit and MOCA (combined with minimum norm estimate) through simulations. SC-MUSIC shows substantial improvement in the localization accuracy compared to these methods. We also show results for real MEG data of a single subject in the resting state. Four sources are localized in the sensorimotor area at f=11Hz which is the expected region for the idle rhythm.

Wedge MUSIC: a novel approach to examine experimental differences of brain source connectivity patterns from EEG/MEG data.

We introduce a novel method to estimate bivariate synchronization, i.e. interacting brain sources at a specific frequency or band, from MEG or EEG data robust to artifacts of volume conduction. The data driven calculation is solely based on the imaginary part of the cross-spectrum as opposed to the imaginary part of coherency. In principle, the method quantifies how strong a synchronization between a distinct pair of brain sources is present in the data. As an input of the method all pairs of pre-defined locations inside the brain can be used which is computationally exhaustive. In contrast to that, reference sources can be used that have been identified by any source reconstruction technique in a prior analysis step. We introduce different variants of the method and evaluate the performance in simulations. As a particular advantage of the proposed methodology, we demonstrate that the novel approach is capable of investigating differences in brain source interactions between experimental conditions or with respect to a certain baseline. For measured data, we first show the application on resting state MEG data where we find locally synchronized sources in the motor-cortex based on the sensorimotor idle rhythms. Finally, we show an example on EEG motor imagery data where we contrast hand and foot movements. Here, we also find local interactions in the expected brain areas.

Univariate normalization of bispectrum using Hölder's inequality.

Considering that many biological systems including the brain are complex non-linear systems, suitable methods capable of detecting these non-linearities are required to study the dynamical properties of these systems. One of these tools is the third order cummulant or cross-bispectrum, which is a measure of interfrequency interactions between three signals. For convenient interpretation, interaction measures are most commonly normalized to be independent of constant scales of the signals such that its absolute values are bounded by one, with this limit reflecting perfect coupling. Although many different normalization factors for cross-bispectra were suggested in the literature these either do not lead to bounded measures or are themselves dependent on the coupling and not only on the scale of the signals. In this paper we suggest a normalization factor which is univariate, i.e., dependent only on the amplitude of each signal and not on the interactions between signals. Using a generalization of Hölder's inequality it is proven that the absolute value of this univariate bicoherence is bounded by zero and one. We compared three widely used normalizations to the univariate normalization concerning the significance of bicoherence values gained from resampling tests. Bicoherence values are calculated from real EEG data recorded in an eyes closed experiment from 10 subjects. The results show slightly more significant values for the univariate normalization but in general, the differences are very small or even vanishing in some subjects. Therefore, we conclude that the normalization factor does not play an important role in the bicoherence values with regard to statistical power, although a univariate normalization is the only normalization factor which fulfills all the required conditions of a proper normalization.

EEG alpha spindles and prolonged brake reaction times during auditory distraction in an on-road driving study.

Driver distraction is responsible for a substantial number of traffic accidents. This paper describes the impact of an auditory secondary task on drivers' mental states during a primary driving task. N=20 participants performed the test procedure in a car following task with repeated forced braking on a non-public test track. Performance measures (provoked reaction time to brake lights) and brain activity (EEG alpha spindles) were analyzed to describe distracted drivers. Further, a classification approach was used to investigate whether alpha spindles can predict drivers' mental states. Results show that reaction times and alpha spindle rate increased with time-on-task. Moreover, brake reaction times and alpha spindle rate were significantly higher while driving with auditory secondary task opposed to driving only. In single-trial classification, a combination of spindle parameters yielded a median classification error of about 8% in discriminating the distracted from the alert driving. Reduced driving performance (i.e., prolonged brake reaction times) during increased cognitive load is assumed to be indicated by EEG alpha spindles, enabling the quantification of driver distraction in experiments on public roads without verbally assessing the drivers' mental states.

Identifying causal networks of neuronal sources from EEG/MEG data with the phase slope index: a simulation study.

The investigation of functional neuronal synchronization has recently become a growing field of research. With high temporal resolution, electroencephalography and magnetoencephalography are well-suited measurement techniques to identify networks of interacting sources underlying the recorded data. The analysis of the data in terms of effective connectivity, nevertheless, contains intrinsic issues such as the problem of volume conduction and the non-uniqueness of the inverse solution. Here, we briefly introduce a series of existing methods assessing these problems. To determine the locations of interacting brain sources robust to volume conduction, all computations are solely based on the imaginary part of the cross-spectrum as a trustworthy source of information. Furthermore, we demonstrate the feasibility of estimating causal relationships of systems of neuronal sources with the phase slope index in realistically simulated data. Finally, advantages and drawbacks of the applied methodology are highlighted and discussed.

Localizing true brain interactions from EEG and MEG data with subspace methods and modified beamformers.

To address the problem of mixing in EEG or MEG connectivity analysis we exploit that noninteracting brain sources do not contribute systematically to the imaginary part of the cross-spectrum. Firstly, we propose to apply the existing subspace method "RAP-MUSIC" to the subspace found from the dominant singular vectors of the imaginary part of the cross-spectrum rather than to the conventionally used covariance matrix. Secondly, to estimate the specific sources interacting with each other, we use a modified LCMV-beamformer approach in which the source direction for each voxel was determined by maximizing the imaginary coherence with respect to a given reference. These two methods are applicable in this form only if the number of interacting sources is even, because odd-dimensional subspaces collapse to even-dimensional ones. Simulations show that (a) RAP-MUSIC based on the imaginary part of the cross-spectrum accurately finds the correct source locations, that (b) conventional RAP-MUSIC fails to do so since it is highly influenced by noninteracting sources, and that (c) the second method correctly identifies those sources which are interacting with the reference. The methods are also applied to real data for a motor paradigm, resulting in the localization of four interacting sources presumably in sensory-motor areas.

Brain Oscillations and Functional Connectivity during Overt Language Production.

In the present study we investigate the communication of different large scale brain sites during an overt language production task with state of the art methods for the estimation of EEG functional connectivity. Participants performed a semantic blocking task in which objects were named in semantically homogeneous blocks of trials consisting of members of a semantic category (e.g., all objects are tools) or in heterogeneous blocks, consisting of unrelated objects. The classic pattern of slower naming times in the homogeneous relative to heterogeneous blocks is assumed to reflect the duration of lexical selection. For the collected data in the homogeneous and heterogeneous conditions the imaginary part of coherency (ImC) was evaluated at different frequencies. The ImC is a measure for detecting the coupling of different brain sites acting on sensor level. Most importantly, the ImC is robust to the artifact of volume conduction. We analyzed the ImC at all pairs of 56 EEG channels across all frequencies. Contrasting the two experimental conditions we found pronounced differences in the theta band at 7 Hz and estimated the most dominant underlying brain sources via a minimum norm inverse solution based on the ImC. As a result of the source localization, we observed connectivity between occipito-temporal and frontal areas, which are well-known to play a major role in lexical-semantic language processes. Our findings demonstrate the feasibility of investigating interactive brain activity during overt language production.

Estimating true brain connectivity from EEG/MEG data invariant to linear and static transformations in sensor space.

The imaginary part of coherency is a measure to investigate the synchronization of brain sources on the EEG/MEG sensor level, robust to artifacts of volume conduction meaning that independent sources cannot generate a significant result. It does not mean, however, that volume conduction is irrelevant when true interactions are present. Here, we analyze in detail the possibilities to construct measures of true brain interactions which are strictly invariant to linear spatial transformations of the sensor data. Specifically, such measures can be constructed from maximization of imaginary coherency in virtual channels, bivariate measures as a corrected variate of imaginary coherence, and global measures indicating the total interaction contained within a space or between two spaces. A complete theoretic framework on this question is provided for second order statistical moments. Relations to existing linear and nonlinear approaches are presented. We applied the methods to resting state EEG data, showing clear interactions at all bands, and to a combined measurement of EEG and MEG during rest condition and a finger tapping task. We found that MEG was capable of observing brain interactions which were not observable in the EEG data.