Search

• Stroke lesions should be modeled in MEG source reconstruction. • Special care for the study of deep sources or asymmetric brain activity. • Strong effects can extend up to 20 mm from the lesion. • Bigger lesions lead to stronger effects, independently from the lesion location. • More priority should be given to usability and accessibility of computational tools. Source reconstruction of magnetoencephalography (MEG) has been used to assess brain reorganization after brain damage, such as stroke. Lesions result in parts of the brain having an electrical conductivity that differs from the normal values. The effect this has on the forward solutions (i.e., the propagation of electric currents and magnetic fields generated by cortical activity) is well predictable. However, their influence on source localization results is not well characterized and understood. This is specifically a concern for patient studies with asymmetric (i.e., within one hemisphere) lesions focusing on asymmetric and lateralized brain activity, such as language. In particular, it is good practice to consider the level of geometrical detail that is necessary to compute and interpret reliable source reconstruction results. To understand the effect of lesions on source estimates and propose recommendations to researchers working with clinical data, in this study we consider the trade off between improved accuracy and the additional effort to compute more realistic head models, with the aim to answer the question whether the additional effort is worth it. We simulated and analyzed the effects of a stroke lesion (i.e., an asymmetrically distributed CSF-filled cavity) in the head model with three different sizes and locations when performing MEG source reconstruction using a finite element method (FEM). We compared the effect of the lesion with a homogeneous head model that neglects the lesion. We computed displacement and attenuation/amplification maps to quantify the localization errors and signal magnitude modulation. We conclude that brain lesions leading to asymmetrically distributed CSF-filled cavities should be modeled when performing MEG source reconstruction, especially when investigating deep sources or post-stroke hemispheric lateralization of functions. The strongest effects are not only visible in perilesional areas, but can extend up to 20 mm from the lesion. Bigger lesions lead to stronger effects impacting larger areas, independently from the lesion location. Lastly, we conclude that more priority should be given to usability and accessibility of the required computational tools, to allow researchers with less technical expertise to use the improved methods that are available but currently not widely adopted yet. [ABSTRACT FROM AUTHOR]

The increasing requirements for adoption of FAIR data management and sharing original research data from neuroimaging studies can be at odds with protecting the anonymity of the research participants due to the person-identifiable anatomical features in the data. We propose a solution to this dilemma for anatomical MRIs used in MEG source analysis. In MEG analysis, the channel-level data is reconstructed to the source-level using models derived from anatomical MRIs. Sharing data, therefore, requires sharing the anatomical MRI to replicate the analysis. The suggested solution is to replace the individual anatomical MRIs with individualised warped templates that can be used to carry out the MEG source analysis and that provide sufficient geometrical similarity to the original participants' MRIs. First, we demonstrate how the individualised template warping can be implemented with one of the leading open-source neuroimaging analysis toolboxes. Second, we compare results from four different MEG source reconstruction methods performed with an individualised warped template to those using the participant's original MRI. While the source reconstruction results are not numerically identical, there is a high similarity between the results for single dipole fits, dynamic imaging of coherent sources beamforming, and atlas-based virtual channel beamforming. There is a moderate similarity between minimum-norm estimates, as anticipated due to this method being anatomically constrained and dependent on the exact morphological features of the cortical sheet. We also compared the morphological features of the warped template to those of the original MRI. These showed a high similarity in grey matter volume and surface area, but a low similarity in the average cortical thickness and the mean folding index within cortical parcels. Taken together, this demonstrates that the results obtained by MEG source reconstruction can be preserved with the warped templates, whereas the anatomical and morphological fingerprint is sufficiently altered to protect the anonymity of research participants. In cases where participants consent to sharing anatomical MRI data, it remains preferable to share the original defaced data with an appropriate data use agreement. In cases where participants did not consent to share their MRIs, the individualised warped MRI template offers a good compromise in sharing data for reuse while retaining anonymity for research participants. [ABSTRACT FROM AUTHOR]

Increased beta (13–30 Hz) oscillatory synchrony in basal ganglia–cortical circuits is a physiological characteristic of Parkinson's disease (PD). While the function of the beta rhythm is unknown, there is evidence that its modulation serves a predictive role, in preparation of future actions. We investigate the relation between predictive beta modulation and entrainment of brain oscillations in a task inviting behavioral entrainment by a regular task structure. MEG was recorded during a serial choice response task, in a group of 12 PD patients and 12 control subjects. In one condition, the reaction stimuli allowed for temporal preparation only ( random condition ), while in a second condition ( predictable condition ) the reaction stimuli allowed both temporal and effector preparation. Reaction times were identical between groups, and both groups benefited equally from the known effector side in the predictable condition. Analysis of oscillatory activity, by contrast, revealed marked differences between groups. In patients, the proportion of preparatory beta power desynchronization preceding the reaction stimuli was significantly smaller than in controls, while the proportion of beta desynchronization following the events was larger. In addition to this shift from prospective to reactive modulation of beta-band oscillations, patients showed a trend to reduced motor cortical pre-stimulus delta phase synchronization, and later gamma power synchronization than controls. Delta phase synchronization was, furthermore, significantly correlated with predictive beta desynchronization, supporting the relevance of hierarchical coupling between oscillations of different frequencies for the analysis of oscillatory changes in PD. Together, these features of task-related oscillatory activity indicate that entrainment fails to engender the same predictive mode of motor activation in PD patients as in healthy controls. [ABSTRACT FROM AUTHOR]

Abstract: A single brief visual stimulus accompanied by two brief tactile stimuli is frequently perceived incorrectly as two flashes, a phenomenon called double-flash illusion (DFI). We investigated whether the DFI is accompanied by changes in rhythmic neuronal activity, using magnetoencephalography in human subjects. Twenty-two subjects received visuo-tactile stimulation and reported the number of perceived visual stimuli. We sorted trials with identical physical stimulation according to the reported subjective percept and assessed differences in spectral power in somatosensory and occipital sensors. In DFI trials, occipital sensors displayed a contralateral enhancement of gamma-band (80–140Hz) activity in response to stimulation. In somatosensory sensors, the DFI was associated with an increase of spectral power for low frequencies (5–17.5Hz) around stimulation and a decrease of spectral power in the 22.5–30Hz range between 450 and 750ms post-stimulation. In summary, several components of rhythmic activity predicted variable subjective experience for constant physical stimulation. Notably, the enhanced occipital gamma-band activity during DFI was similar in time and frequency extent to the somatosensory gamma-band response to tactile stimulation. We speculate that the DFI might therefore occur when the somatosensory gamma-response is transmitted to visual cortex. This transmission might be supported by the observed modulations in low-frequency activity. [ABSTRACT FROM AUTHOR]

Abstract: The current state of empirical investigations refers to consciousness as an all-or-none phenomenon. However, a recent theoretical account opens up this perspective by proposing a partial level (between nil and full) of conscious perception. In the well-studied case of single-word reading, short-lived exposure can trigger incomplete word-form recognition wherein letters fall short of forming a whole word in one's conscious perception thereby hindering word-meaning access and report. Hence, the processing from incomplete to complete word-form recognition straightforwardly mirrors a transition from partial to full-blown consciousness. We therefore hypothesized that this putative functional bottleneck to consciousness (i.e. the perceptual boundary between partial and full conscious perception) would emerge at a major key hub region for word-form recognition during reading, namely the left occipito-temporal junction. We applied a real-time staircase procedure and titrated subjective reports at the threshold between partial (letters) and full (whole word) conscious perception. This experimental approach allowed us to collect trials with identical physical stimulation, yet reflecting distinct perceptual experience levels. Oscillatory brain activity was monitored with magnetoencephalography and revealed that the transition from partial-to-full word-form perception was accompanied by alpha-band (7–11Hz) power suppression in the posterior left occipito-temporal cortex. This modulation of rhythmic activity extended anteriorly towards the visual word form area (VWFA), a region whose selectivity for word-forms in perception is highly debated. The current findings provide electrophysiological evidence for a functional bottleneck to consciousness thereby empirically instantiating a recently proposed partial perspective on consciousness. Moreover, the findings provide an entirely new outlook on the functioning of the VWFA as a late bottleneck to full-blown conscious word-form perception. [Copyright &y& Elsevier]

Abstract: Groups of activated neurons typically synchronize in the gamma-frequency band (30-100Hz), and gamma-band synchronization has been implicated in numerous cognitive functions. Those functions are ultimately expressed as behavior and therefore, functional gamma-band synchronization should be directly related to behavior. We recorded the magnetoencephalogram in human subjects and used a visual stimulus to induce occipital gamma-band activity. We found that the strength of this gamma-band activity at a given moment predicted the speed with which the subject was able to report a change in the stimulus. This predictive effect was restricted in time, frequency and space: It started only around 200ms before the behaviorally relevant stimulus change, was present only between 50 and 80Hz, and was significant only in bilateral middle occipital gyrus, while the peak of overall visually induced gamma-band activity was found in the calcarine sulcus. These results suggest that visually induced gamma-band activity is functionally relevant for the efficient transmission of stimulus change information to brain regions issuing the corresponding motor response. [Copyright &y& Elsevier]

• Concise overview and explanation of beamformers for M/EEG data analysis. • Practical considerations and best practices for beamforming analyses. • Unification of terminology across popular open source software packages. • Comparison of implementations and user interfaces between software packages. Beamforming is a popular method for functional source reconstruction using magnetoencephalography (MEG) and electroencephalography (EEG) data. Beamformers, which were first proposed for MEG more than two decades ago, have since been applied in hundreds of studies, demonstrating that they are a versatile and robust tool for neuroscience. However, certain characteristics of beamformers remain somewhat elusive and there currently does not exist a unified documentation of the mathematical underpinnings and computational subtleties of beamformers as implemented in the most widely used academic open source software packages for MEG analysis (Brainstorm, FieldTrip, MNE, and SPM). Here, we provide such documentation that aims at providing the mathematical background of beamforming and unifying the terminology. Beamformer implementations are compared across toolboxes and pitfalls of beamforming analyses are discussed. Specifically, we provide details on handling rank deficient covariance matrices, prewhitening, the rank reduction of forward fields, and on the combination of heterogeneous sensor types, such as magnetometers and gradiometers. The overall aim of this paper is to contribute to contemporary efforts towards higher levels of computational transparency in functional neuroimaging. [ABSTRACT FROM AUTHOR]

Abstract: Magnetoencephalographic (MEG) recordings are a rich source of information about the neural dynamics underlying cognitive processes in the brain, with excellent temporal and good spatial resolution. In recent years there have been considerable advances in MEG hardware developments and methods. Sophisticated analysis techniques are now routinely applied and continuously improved, leading to fascinating insights into the intricate dynamics of neural processes. However, the rapidly increasing level of complexity of the different steps in a MEG study make it difficult for novices, and sometimes even for experts, to stay aware of possible limitations and caveats. Furthermore, the complexity of MEG data acquisition and data analysis requires special attention when describing MEG studies in publications, in order to facilitate interpretation and reproduction of the results. This manuscript aims at making recommendations for a number of important data acquisition and data analysis steps and suggests details that should be specified in manuscripts reporting MEG studies. These recommendations will hopefully serve as guidelines that help to strengthen the position of the MEG research community within the field of neuroscience, and may foster discussion in order to further enhance the quality and impact of MEG research. [Copyright &y& Elsevier]

Beamformers are applied for estimating spatiotemporal characteristics of neuronal sources underlying measured MEG/EEG signals. Several MEG analysis toolboxes include an implementation of a linearly constrained minimum-variance (LCMV) beamformer. However, differences in implementations and in their results complicate the selection and application of beamformers and may hinder their wider adoption in research and clinical use. Additionally, combinations of different MEG sensor types (such as magnetometers and planar gradiometers) and application of preprocessing methods for interference suppression, such as signal space separation (SSS), can affect the results in different ways for different implementations. So far, a systematic evaluation of the different implementations has not been performed. Here, we compared the localization performance of the LCMV beamformer pipelines in four widely used open-source toolboxes (MNE-Python, FieldTrip, DAiSS (SPM12), and Brainstorm) using datasets both with and without SSS interference suppression. We analyzed MEG data that were i) simulated, ii) recorded from a static and moving phantom, and iii) recorded from a healthy volunteer receiving auditory, visual, and somatosensory stimulation. We also investigated the effects of SSS and the combination of the magnetometer and gradiometer signals. We quantified how localization error and point-spread volume vary with the signal-to-noise ratio (SNR) in all four toolboxes. When applied carefully to MEG data with a typical SNR (3–15 ​dB), all four toolboxes localized the sources reliably; however, they differed in their sensitivity to preprocessing parameters. As expected, localizations were highly unreliable at very low SNR, but we found high localization error also at very high SNRs for the first three toolboxes while Brainstorm showed greater robustness but with lower spatial resolution. We also found that the SNR improvement offered by SSS led to more accurate localization. • Different beamformer implementations are reported to sometimes yield differing source estimates for the same MEG data. • We compared beamformers in four major open-source MEG analysis toolboxes. • All toolboxes provide consistent and accurate results with 3–15-dB input SNR. • However, localization errors are high at very high input SNR for the tested scalar beamformers. • We discuss the critical differences between the implementations. [ABSTRACT FROM AUTHOR]