Search

Machine learning (ML) is increasingly used in cognitive, computational and clinical neuroscience. The reliable and efficient application of ML requires a sound understanding of its subtleties and limitations. Training ML models on datasets with imbalanced classes is a particularly common problem, and it can have severe consequences if not adequately addressed. With the neuroscience ML user in mind, this paper provides a didactic assessment of the class imbalance problem and illustrates its impact through systematic manipulation of data imbalance ratios in (i) simulated data and (ii) brain data recorded with electroencephalography (EEG) and magnetoencephalography (MEG). Our results illustrate how the widely-used Accuracy (Acc) metric, which measures the overall proportion of successful predictions, yields misleadingly high performances, as class imbalance increases. Because Acc weights the per-class ratios of correct predictions proportionally to class size, it largely disregards the performance on the minority class. A binary classification model that learns to systematically vote for the majority class will yield an artificially high decoding accuracy that directly reflects the imbalance between the two classes, rather than any genuine generalizable ability to discriminate between them. We show that other evaluation metrics such as the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC), and the less common Balanced Accuracy (BAcc) metric – defined as the arithmetic mean between sensitivity and specificity, provide more reliable performance evaluations for imbalanced data. Our findings also highlight the robustness of Random Forest (RF), and the benefits of using stratified cross-validation and hyperprameter optimization to tackle data imbalance. Critically, for neuroscience ML applications that seek to minimize overall classification error, we recommend the routine use of BAcc, which in the specific case of balanced data is equivalent to using standard Acc, and readily extends to multi-class settings. Importantly, we present a list of recommendations for dealing with imbalanced data, as well as open-source code to allow the neuroscience community to replicate and extend our observations and explore alternative approaches to coping with imbalanced data.

We present a standalone Matlab software platform complete with visualization for the reconstruction of the neural activity in the brain from MEG or EEG data. The underlying inversion combines hierarchical Bayesian models and Krylov subspace iterative least squares solvers. The Bayesian framework of the underlying inversion algorithm allows to account for anatomical information and possible a priori belief about the focality of the reconstruction. The computational efficiency makes the software suitable for the reconstruction of lengthy time series on standard computing equipment. The algorithm requires minimal user provided input parameters, although the user can express the desired focality and accuracy of the solution. The code has been designed so as to favor the parallelization performed automatically by Matlab, according to the resources of the host computer. We demonstrate the flexibility of the platform by reconstructing activity patterns with supports of different sizes from MEG and EEG data. Moreover, we show that the software reconstructs well activity patches located either in the subcortical brain structures or on the cortex. The inverse solver and visualization modules can be used either individually or in combination. We also provide a version of the inverse solver that can be used within Brainstorm toolbox. All the software is available online by Github, including the Brainstorm plugin, with accompanying documentation and test data.

Schizophrenia has a complex etiology and symptomatology that is difficult to untangle. After decades of research, important advancements towards a central biomarker are still lacking. One of the missing pieces is a better understanding of how non-linear neural dynamics are altered in this patient population. In this study, the resting-state neuromagnetic signals of schizophrenia patients and healthy controls were analyzed in the framework of criticality. When biological systems like the brain are in a state of criticality, they are thought to be functioning at maximum efficiency (e.g., optimal communication and storage of information) and with maximum adaptability to incoming information. Here, we assessed the self-similarity and multifractality of resting-state brain signals recorded with magnetoencephalography in patients with schizophrenia patients and in matched controls. Our analysis showed a clear ascending, rostral to caudal gradient of self-similarity values in healthy controls, and an opposite gradient for multifractality (descending values, rostral to caudal). Schizophrenia patients had similar, although attenuated, gradients of self-similarity and multifractality values. Statistical tests showed that patients had higher values of self-similarity than controls in fronto-temporal regions, indicative of more regularity and memory in the signal. In contrast, patients had less multifractality than controls in the parietal and occipital regions, indicative of less diverse singularities and reduced variability in the signal. In addition, supervised machine-learning, based on logistic regression, successfully discriminated the two groups using measures of self-similarity and multifractality as features. Our results provide new insights into the baseline cognitive functioning of schizophrenia patients by identifying key alterations of criticality properties in their resting-state brain data.

Electrical source imaging (ESI) aims at reconstructing the electrical brain activity from measurements of the electric field on the scalp. Even though the localization of single focal sources should be relatively straightforward, different methods provide diverse solutions due to the different underlying assumptions. Furthermore, their input parameter(s) further affects the solution provided by each method, making localization even more challenging. In addition, validations and comparisons are typically performed either on synthetic data or through post-operative outcomes, in both cases with considerable limitations.We use an in-vivo high-density EEG dataset recorded during intracranial single pulse electrical stimulation, in which the true sources are substantially dipolar and their locations are known. We compare ten different ESI methods under multiple choices of input parameters, to assess the accuracy of the best reconstruction, as well as the impact of the parameters on the localization performance.Best reconstructions often fall within 1 cm from the true source, with more accurate methods outperforming less accurate ones by 1 cm, on average. Expectedly, dipolar methods tend to outperform distributed methods. Sensitivity to input parameters varies widely between methods. Depth weighting played no role for three out of six methods implementing it. In terms of regularization parameters, for several distributed methods SNR=1 unexpectedly turned out to be the best choice among the tested ones.Our data show similar levels of accuracy of ESI techniques when applied to “conventional” (32 channels) and dense (64, 128, 256 channels) EEG recordings.Overall findings reinforce the importance that ESI may have in the clinical context, especially when applied to identify the surgical target in potential candidates for epilepsy surgery.

The MagnetoEncephaloGraphy (MEG) has gained great interest in neurorehabilitation training due to its high temporal resolution. The challenge is to localize the active regions of the brain in a fast and accurate way. In this paper we use an inversion method based on random spatial sampling to solve the real-time MEG inverse problem. Several numerical tests on synthetic but realistic data show that the method takes just a few hundredths of a second on a laptop to produce an accurate map of the electric activity inside the brain. Moreover, it requires very little memory storage. For this reasons the random sampling method is particularly attractive in real-time MEG applications., 15 pages, 15 figures

MagnetoEncephaloGraphy (MEG) and ElectroEncephaloGraphy (EEG) are the most common non-invasive brain imaging techniques for monitoring the electrical brain activity with millisecond resolution. Due to their high, millisecond, temporal resolution, these techniques are also the most suitable for studying the dynamic interplay of between brain regions during information processing. In clinical settings, MEG and EEG are valuable methods for the pre-surgical evaluation of patients with pharmaco-resistant epilepsy. The localization of epileptogenic zone (EZ) provided by these techniques can avoid or guide invasive intracranial recordings. More recently, there is has been a growing interest in exploiting leveraging these techniques as a crucial tool to advance our understanding of the pathophysiological mechanisms of major neuropsychiatric disorders and neurological diseases. This symposium offers an overview of very recent studies on the use ofwhich employed MEG and EEG as valid tools to investigate brain disorders. In particularSpecifically, the five talks will provide important insights on new methodological approaches to characterize brain dysfunctions in different pathological conditions. We will focus on : (1) emerging tool to characterize irritative zones in epileptic patients and to detect in a semi automatic way epileptiform abnormalities during EEG sleep recordings ina semi-automatic way; (2) two different methodological approaches to evaluate frequency specific alterations of brain activity and connectivity in schizoprenia and glioma patients; and (3) using machine-learning methods for identificationto identify EEG predictors of dementia in Parkinson.

The present paper represents a first methodological work for the construction of a robust and accurate algorithm for the solution of an inverse problem given by the identification of the parameters of a lumped mathematical model of fetal circulation introduced by G. Pennati et al. (1997). The underlying estimation techniques here applied are two global search meth- ods, respectively a Parameter Space Investigation (PSI) and the Ensemble Kalman Filter (EnKF), with a refinement performed with a local search method, i.e. Levenberg- Marquardt method (LM). The results here presented show the soundness of our methodology and opens the possibility to apply these techniques for the parameter identification of waveforms obtained from Doppler clinical measurements in the next future. Our final goal is to build a non-invasive simulation tool for the description of the circulation of fetuses in the context of a patient-specific model in order to help clinicians in early diagnosis of pathologies like cardiac distress or growth retardation.

Meditation practices have been claimed to have a positive effect on the regulation of mood and emotion for quite some time by practitioners, and in recent times there has been a sustained effort to provide a more precise description of the changes induced by meditation on human brain. Longitudinal studies have reported morphological changes in cortical thickness and volume in selected brain regions due to meditation practice, which is interpreted as evidence for effectiveness of it beyond the subjective self reporting. Evidence based on real time monitoring of meditating brain by functional imaging modalities such as MEG or EEG remains a challenge. In this article we consider MEG data collected during meditation sessions of experienced Buddhist monks practicing focused attention (Samatha) and open monitoring (Vipassana) meditation, contrasted by resting state with eyes closed. The MEG data is first mapped to time series of brain activity averaged over brain regions corresponding to a standard Destrieux brain atlas, and further by bootstrapping and spectral analysis to data matrices representing a random sample of power spectral densities over bandwidths corresponding to $\alpha$, $\beta$, $\gamma$, and $\theta$ bands in the spectral range. We demonstrate using linear discriminant analysis (LDA) that the samples corresponding to different meditative or resting states contain enough fingerprints of the brain state to allow a separation between different states, and we identify the brain regions that appear to contribute to the separation. Our findings suggest that cingulate cortex, insular cortex and some of the internal structures, most notably accumbens, caudate and putamen nuclei, thalamus and amygdalae stand out as separating regions, which seems to correlate well with earlier findings based on longitudinal studies., Comment: 34 pages, 15 figures

Schizophrenia has a complex etiology and symptomatology that is difficult to untangle. After decades of research, important advancements toward a central biomarker are still lacking. One of the missing pieces is a better understanding of how non-linear neural dynamics are altered in this patient population. In this study, the resting-state neuromagnetic signals of schizophrenia patients and healthy controls were analyzed in the framework of criticality. When biological systems like the brain are in a state of criticality, they are thought to be functioning at maximum efficiency (e.g., optimal communication and storage of information) and with maximum adaptability to incoming information. Here, we assessed the self-similarity and multifractality of resting-state brain signals recorded with magnetoencephalography in patients with schizophrenia patients and in matched controls. Schizophrenia patients had similar, although attenuated, patterns of self-similarity and multifractality values. Statistical tests showed that patients had higher values of self-similarity than controls in fronto-temporal regions, indicative of more regularity and memory in the signal. In contrast, patients had less multifractality than controls in the parietal and occipital regions, indicative of less diverse singularities and reduced variability in the signal. In addition, supervised machine-learning, based on logistic regression, successfully discriminated the two groups using measures of self-similarity and multifractality as features. Our results provide new insights into the baseline cognitive functioning of schizophrenia patients by identifying key alterations of criticality properties in their resting-state brain data.

Highlights • Temporal autocorrelation is attenuated in the neural dynamics of schizophrenia. • Alterations are primarily observed in the alpha and beta frequency bands. • Machine-learning significantly classified patients and controls with 82% accuracy. • Brain regions showing disrupted temporal connectivity include (para)limbic areas. • Long-range temporal correlation informs on temporal dysconnectivity in schizophrenia., Current theories of schizophrenia emphasize the role of altered information integration as the core dysfunction of this illness. While ample neuroimaging evidence for such accounts comes from investigations of spatial connectivity, understanding temporal disruptions is important to fully capture the essence of dysconnectivity in schizophrenia. Recent electrophysiology studies suggest that long-range temporal correlation (LRTC) in the amplitude dynamics of neural oscillations captures the integrity of transferred information in the healthy brain. Thus, in this study, 25 schizophrenia patients and 25 controls (8 females/group) were recorded during two five-minutes of resting-state magnetoencephalography (once with eyes-open and once with eyes-closed). We used source-level analyses to investigate temporal dysconnectivity in patients by characterizing LRTCs across cortical and sub-cortical brain regions. In addition to standard statistical assessments, we applied a machine learning framework using support vector machine to evaluate the discriminative power of LRTCs in identifying patients from healthy controls. We found that neural oscillations in schizophrenia patients were characterized by reduced signal memory and higher variability across time, as evidenced by cortical and subcortical attenuations of LRTCs in the alpha and beta frequency bands. Support vector machine significantly classified participants using LRTCs in key limbic and paralimbic brain areas, with decoding accuracy reaching 82%. Importantly, these brain regions belong to networks that are highly relevant to the symptomology of schizophrenia. These findings thus posit temporal dysconnectivity as a hallmark of altered information processing in schizophrenia, and help advance our understanding of this pathology.

Recent years have witnessed a massive push towards reproducible research in neuroscience. Unfortunately, this endeavor is often challenged by the large diversity of tools used, project-specific custom code and the difficulty to track all user-defined parameters. NeuroPycon is an open-source multi-modal brain data analysis toolkit which provides Python-based template pipelines for advanced multi-processing of MEG, EEG, functional and anatomical MRI data, with a focus on connectivity and graph theoretical analyses. Importantly, it provides shareable parameter files to facilitate replication of all analysis steps. NeuroPycon is based on the NiPype framework which facilitates data analyses by wrapping many commonly-used neuroimaging software tools into a common Python environment. In other words, rather than being a brain imaging software with is own implementation of standard algorithms for brain signal processing, NeuroPycon seamlessly integrates existing packages (coded in python, Matlab or other languages) into a unified python framework. Importantly, thanks to the multi-threaded processing and computational efficiency afforded by NiPype, NeuroPycon provides an easy option for fast parallel processing, which critical when handling large sets of multi-dimensional brain data. Moreover, its flexible design allows users to easily configure analysis pipelines by connecting distinct nodes to each other. Each node can be a Python-wrapped module, a user-defined function or a well-established tool (e.g. MNE-Python for MEG analysis, Radatools for graph theoretical metrics, etc.). Last but not least, the ability to use NeuroPycon parameter files to fully describe any pipeline is an important feature for reproducibility, as they can be shared and used for easy replication by others. The current implementation of NeuroPycon contains two complementary packages: The first, called ephypype, includes pipelines for electrophysiology analysis and a command-line interface for on the fly pipeline creation. Current implementations allow for MEG/EEG data import, pre-processing and cleaning by automatic removal of ocular and cardiac artefacts, in addition to sensor or source-level connectivity analyses. The second package, called graphpype, is designed to investigate functional connectivity via a wide range of graph-theoretical metrics, including modular partitions. The present article describes the philosophy, architecture, and functionalities of the toolkit and provides illustrative examples through interactive notebooks. NeuroPycon is available for download via github (https://github.com/neuropycon) and the two principal packages are documented online (https://neuropycon.github.io/ephypype/index.html. and https://neuropycon.github.io/graphpype/index.html). Future developments include fusion of multi-modal data (eg. MEG and fMRI or intracranial EEG and fMRI). We hope that the release of NeuroPycon will attract many users and new contributors, and facilitate the efforts of our community towards open source tool sharing and development, as well as scientific reproducibility.

Despite numerous important contributions, the investigation of brain connectivity with magnetoencephalography (MEG) still faces multiple challenges. One critical aspect of source-level connectivity, largely overlooked in the literature, is the putative effect of the choice of the inverse method on the subsequent cortico-cortical coupling analysis. We set out to investigate the impact of three inverse methods on source coherence detection using simulated MEG data. To this end, thousands of randomly located pairs of sources were created. Several parameters were manipulated, including inter- and intra-source correlation strength, source size and spatial configuration. The simulated pairs of sources were then used to generate sensor-level MEG measurements at varying signal-to-noise ratios (SNR). Next, the source level power and coherence maps were calculated using three methods (a) L2-Minimum-Norm Estimate (MNE), (b) Linearly Constrained Minimum Variance (LCMV) beamforming, and (c) Dynamic Imaging of Coherent Sources (DICS) beamforming. The performances of the methods were evaluated using Receiver Operating Characteristic (ROC) curves. The results indicate that beamformers perform better than MNE for coherence reconstructions if the interacting cortical sources consist of point-like sources. On the other hand, MNE provides better connectivity estimation than beamformers, if the interacting sources are simulated as extended cortical patches, where each patch consists of dipoles with identical time series (high intra-patch coherence). However, the performance of the beamformers for interacting patches improves substantially if each patch of active cortex is simulated with only partly coherent time series (partial intra-patch coherence). These results demonstrate that the choice of the inverse method impacts the results of MEG source-space coherence analysis, and that the optimal choice of the inverse solution depends on the spatial and synchronization profile of the interacting cortical sources. The insights revealed here can guide method selection and help improve data interpretation regarding MEG connectivity estimation.

Magnetoencephalography (MEG) aims at reconstructing the unknown neuroelectric activity in the brain from non-invasive measurements of the magnetic field induced by neural sources. The solution of this ill-posed, ill-conditioned inverse problem is usually dealt with using regularization techniques that are often time-consuming, and computationally and memory storage demanding. In this paper we analyze how a slimmer procedure, random sampling, affects the estimation of the brain activity generated by both synthetic and real sources.

Biological systems tend to display complex behaviour with a power-law (1/f - like) distribution. In the brain, this translates into neural activity that exhibits scale-free, temporal or spatial, properties (He, 2014). Scaleinvariance has been observed across different neuroimaging modalities and conditions (Linkenkaer-Hansen, 2001; He, 2014; Ciuciu et al. 2012). Beyond previously used features, recent electrophysiology studies have shown the presence of long-range temporal correlations (LRTCs) in the amplitude dynamics of alpha and beta oscillations (Nikulin et al. 2012). Disease, such as psychosis, can alter the temporal properties of neuronal activity and, consequently, potentially affect information integration (Fernandez et al. 2013). The goals of this study were to: (a) measure and compare scale-free dynamics in schizophrenia patients (SZ) and controls using magnetoencephalography (MEG), and (b) classify subjects using machine-learning tools. Five minutes of resting-state MEG were acquired for 25 SZ patients and 25 controls during eyes open and eyes closed conditions. Detrended Fluctuation analysis (DFA) was applied to resting alpha and beta band oscillatory amplitudes to investigate LRTCs in each subject group, at both the sensor and the source levels. Permutation tests with maximum statistics correction were used to explore group differences (p < 0.01). Finally, machine-learning, using support vector machine (SVM) and a 10-fold cross-validation technique, was applied to classify controls and patients based on DFA values. Statistically significant decoding was assessed using binomial law statistics and permutation tests (p < 0.001). Results/Discussion: Significant group differences were observed between the two groups at both sensor and source levels. Specifically, sensormeasured DFA were found to be significantly attenuated in SZ patients compared to controls over the temporo-parietal areas in the alpha-band, and over central brain regions in the beta-band. Source-level analysis improved anatomical. Relative group differences in DFA were found up to 20% in both alpha and beta bands. Specifically, compared to controls, attenuated DFA values were observed in the frontal, temporal and occipital poles and the cuneus in the alpha band, and in the occipital pole, cuneus and mid frontal gyrus in the beta band. Finally, the machine-learning algorithm successfully classified the groups using the measure of DFA with up to 76% decoding accuracy. The combination of classical statistical measures and machine learning tools in our study illustrate the interest of using features of scale-free dynamics to enhance our understanding of schizophrenia and potentially find a new path for early clinical diagnosis.

A recently proposed iterated alternating sequential (IAS) MEG inverse solver algorithm, based on the coupling of a hierarchical Bayesian model with computationally efficient Krylov subspace linear solver, has been shown to perform well for both superficial and deep brain sources. However, a systematic study of its ability to correctly identify active brain regions is still missing. We propose novel statistical protocols to quantify the performance of MEG inverse solvers, focusing in particular on how their accuracy and precision at identifying active brain regions. We use these protocols for a systematic study of the performance of the IAS MEG inverse solver, comparing it with three standard inversion methods, wMNE, dSPM, and sLORETA. To avoid the bias of anecdotal tests towards a particular algorithm, the proposed protocols are Monte Carlo sampling based, generating an ensemble of activity patches in each brain region identified in a given atlas. The performance in correctly identifying the active areas is measured by how much, on average, the reconstructed activity is concentrated in the brain region of the simulated active patch. The analysis is based on Bayes factors, interpreting the estimated current activity as data for testing the hypothesis that the active brain region is correctly identified, versus the hypothesis of any erroneous attribution. The methodology allows the presence of a single or several simultaneous activity regions, without assuming that the number of active regions is known. The testing protocols suggest that the IAS solver performs well with both with cortical and subcortical activity estimation.

Background. With the exponential increase in data dimension and methodological complexities, brain networks analysis with MEG and EEG has become an increasingly challenging and time-consuming endeavor. To date, performing all the data processing steps that are required for a complete MEG/EEG analysis pipeline often require the use of a multitude of software packages and in-house or custom tools (e.g. MRI segmentation, pre-processing, source reconstruction, graph theoretical analysis, statistics). This is not only cumbersome, but may also increase sources of errors and hinders replication of results. Here we describe NeuroPycon, an open-source, multi-modal brain data analysis kit which provides Python-based pipelines for advanced multi-thread processing of fMRI, MEG, and EEG data, with a focus on connectivity and graph analyses [1]. Methods. NeuroPycon is based on the NiPype framework [2] which facilitates data analyses by wrapping numerous commonly-used neuroimaging software solutions into a common python framework. NeuroPycon allows accessing and interfacing with the existing open-science neuroimaging software and signal processing toolboxes, within a unified framework relying on several freely available Python packages which are developed for efficient and fast parallel processing. The current implementation of NeuroPycon comprises three different packages: 2 of 3 - ephypype is mainly based on MNE-Python package [3] and includes pipelines for electrophysiology analyses. Current implementation features MEG/EEG data import, data pre-processing and cleaning via an automatic removal of eyes and heart-related artefacts, and sensor or source-level connectivity analyses - graphpype is based on radatools [4], a set of freely distributed applications aimed at analyses of Complex Networks. It comprises pipelines for functional connectivity studies which heavily exploit graph-theoretical metrics including among other things modular partitions - neuropycon_cli is a command line interface for the ephypype package. Notably, NeuroPycon pipelines can be used in a stand-alone mode but they can also be combined within building blocks to form a larger workflow, in which case the input of one pipeline comes from the outputs of the others. Each pipeline, based upon the nipype engine, is defined by connecting different nodes, with each node being either a user-defined function or a python-wrapped external routine (as MNE-python modules or radatools functions). Results and Discussion. NeuroPycon provides a common and fast framework to develop workflows for advanced neuroimaging data analyses. Several workflows have already been developed to analyze different datasets coming from MEG and EEG studies, such as EEG sleep data and MEG resting state measurements. Furthermore, pipelines defined in graphpype have already been used to perform graph theoretical analysis on a different fMRI datasets. Results visualisation for NeuroPycon is provided through the visbrain (http://visbrain.org/), an open-source multi-purpose python software devoted to graphical representation of neuroscientific data and built on top of VisPy [5], a high-performance visualization library leveraging GPU acceleration. NeuroPycon will shortly be available for download via github (installation via Docker) and is currently being documented (https://neuropycon.github.io/neuropycon_doc/). Future developments will include fusion of multi-modal data (ex. MEG and fMRI or iEEG and fMRI) and feature an increased compatibility with the existing Python packages of interest such as machine learning tools. References: 1. Bullmore E, Sporns O (2009), Nat Rev Neuroscience 2. Gorgolewski et al. (2011) Front. Neuroinformatics 3. Gramfort et al. (2013), Front. Neuroscience 4. http://deim.urv.cat/~sergio.gomez/radatools.php 5. Campagnola et al. (2015), Proceedings of the 14th Python in Science Conference

Introduction : Neuroimaging studies have shown that in adults, the motor learning induced alterations of the functional connectivity assessed during Resting State Networks (RSN) is age-dependent (Mary et al., 2017). Motor learning relies on the build-up of new sensori-motor representations, which has been studied using the bar-man task in adults (Barlaam, Vaugoyeau, Fortin, Assaiante, & Schmitz, 2016; Paulignan, Dufossé, Hugon, & Massion, 1989) and in children (Schmitz et al, 2002).. The aim of this study was to investigate the modulations of functional connectivity after a motor learning task in the childs resting state network. Method : 20 children aged 7 to 12 (12 boys; age 9y 9m ; age 1y et 8m) took part in the study. The resting state tasks consisted of a 3 session closed-eyes recording of 45 seconds. The first one was used as a baseline and the next two have been set around the motor learning task to evaluate the effect of motor learning on the connectivity in the RSN. The motor learning task was a load-lifting task where the participant was asked to lift a weight using its right hand which triggered the fall of a weight attached to the supporting left arm (Paulignan et al, 1989). We continuously recorded the neuromagnetic signals using a 275 channels CTFMEG system. To quantify the functional connectivity between brain regions, coherency analyses have been conducted, using the imaginary part of the coherency, corresponding to the correlation coefficient between two signals in the alpha and beta frequency bands. Moreover, graph theory analysis has provided an overview of the network organisation after the motor task. All resultshave been analysed using data-related permutation statistic with a 0.005 significance threshold (calculated as a relation between the number of conditions (n=2) and the number of subjects (N=20); = 1/, therefore < 0.005). 2 of 3 Results : The motor learning behavioural performances were assessed using a learning curve model throughout the trials which revealed a significant global learning effect (F(7 ;19) = 50,62 ; p

Objective: Neuroimaging studies provide evidence of disturbed resting-state brain networks in Schizophrenia (SZ). However, untangling the neuronal mechanisms that subserve these baseline alterations requires measurement of their electrophysiological underpinnings. This systematic review specifically investigates the contributions of resting-state Magnetoencephalography (MEG) in elucidating abnormal neural organization in SZ patients.& para;& para;Method: A systematic literature review of resting-state MEG studies in SZ was conducted. This literature is discussed in relation to findings from resting-state fMRI and EEG, as well as to task-based MEG research in SZ population. Importantly, methodological limitations are considered and recommendations to overcome current limitations are proposed.& para;& para;Results: Resting-state MEG literature in SZ points towards altered local and long-range oscillatory network dynamics in various frequency bands. Critical methodological challenges with respect to experiment design, and data collection and analysis need to be taken into consideration.& para;& para;Conclusion: Spontaneous MEG data show that local and global neural organization is altered in SZ patients. MEG is a highly promising tool to fill in knowledge gaps about the neurophysiology of SZ. However, to reach its fullest potential, basic methodological challenges need to be overcome.& para;& para;Significance: MEG-based resting-state power and connectivity findings could be great assets to clinical and translational research in psychiatry, and SZ in particular. (C) 2017 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.

An important process underlying meditation and its benefits involves the regulation of attention. Although the two main meditation categories - open-monitoring meditation (OMM) and focused-attention meditation (FAM) - are associated with different benefits and attentional processes, direct comparisons between the attentional neural mechanism of FAM and OMM are rare. This study uses magnetoencephalography (MEG) recordings in 12 expert meditators to compare FAM and OMM by assessing (i) source spectral power, (ii) seed-based functional connectivity of key regions in attention, (including anterior cingulate cortex, dorsolateral prefrontal cortex and the thalamus) and (iii) graph theory metrics that describe brain-wide efficiency of information processing. We reconstructed the source space using minimum norm estimate and computed spectral power and functional connectivity in multiple frequency bands (delta, theta, alpha, beta, gamma) using a custom-designed python-based MEG analysis pipeline (NeuroPycon). The results reveal unique patterns of neural processes specific to FAM or OMM. Among other things, compared to FAM, OMM appears to be characterized by enhanced small-world network properties. By contrast, FAM exhibits greater functional connectivity between the anterior cingulate cortex and frontal regions. These findings shed light onto the mechanisms that potentially mediate the different behavioral and attentional capacities associated with each of the two meditation techniques. Our results are discussed in the context of previous behavioral and fMRI studies on meditation and attention.

NeuroPycon is an open-source multi-modal brain data analysis kit which provides Python-based pipelines for advanced multi-thread processing of fMRI, MEG and EEG data, with a focus on connectivity and graph analyses [1]. NeuroPycon is based on NiPype framework [2] which facilitates data analyses by wrapping many commonly-used neuroimaging software into a common python framework. Therefore, a major strength of NeuroPycon is that it relies on (and interfaces with) several freely available Python packages developed for efficient and fast parallel processing and that it seamlessly connects with existing open-science neuroimaging and signal processing toolboxes. The flexible design allows users to configure analysis pipelines defined by connecting different nodes, where each node may be a user-defined function or a well-established tool or python-wrapped module (e.g. MNE-python for MEG analysis [3], etc.). The current implementation of NeuroPycon contains three different packages: - ephypype includes pipelines for electrophysiology analysis; current implementations allow for MEG/EEG data import, data pre-processing and cleaning by an automatic removal of eyes and heart related artefacts, sensor or source-level connectivity analyses - graphpype allows to study functional connectivity exploiting graph-theoretical metrics including also modular partitions - clipype is a command line interface for ephypype package. NeuroPycon will shortly be available for download via github (installation via Docker) and is currently being documented. Future developments include fusion of multi-modal data (ex. MEG and fMRI or iEEG and fMRI). References 1. Bullmore, Sporns (2009), Nat Rev Neurosci 2. Gorgolewski et al. (2011) Front. Neuroinform 3. Gramfort et al. (2013), Front. Neurosci

Background Electrocorticography (ECoG) measures the distribution of the electrical potentials on the cortex produced by the neural currents. A full interpretation of ECoG data requires solving the ill-posed inverse problem of reconstructing the spatio-temporal distribution of the neural currents. This study addresses the ECoG source modeling developing a beamformer method. New method We computed the lead-field matrix by using a novel routine provided by the OpenMEEG software. We performed an analysis of the numerical stability of the ECoG inverse problem by computing the condition number of the lead-field matrix for different configurations of the electrodes grid. We applied a Linear Constraint Minimum Variance (LCMV) beamformer to both synthetic data and a set of real measurements recorded during a rapid visual categorization task. Results For all considered grids the condition number indicates that the ECoG inverse problem is mildly ill-conditioned. For realistic SNR we found a good performance of the LCMV algorithm for both localization and waveforms reconstruction. Comparison with existing method The flow of information reconstructed by analyzing real data seems consistent with both invasive monkey electrophysiology studies and non-invasive (MEG and fMRI) human studies. Conclusions Despite a growing interest from the neuroscientific community, solving the ECoG inverse problem has not quite yet reached the level of systematicity found for EEG and MEG. Starting from an analysis of the numerical stability of the problem we considered the most widely utilized method for modeling neurophysiological data based on the beamformer method in the hope to establish benchmarks for future studies.

The phenomenologyand reported effects of meditation vary according to the technique practiced.While numerous studies have explored the cerebral mechanisms involved inmeditation, little research provides direct comparisons between the neuronalnetwork dynamics involved in different meditation techniques. Here, we exploreand compare brain signals recorded with magnetoencephalography (MEG) during (a)resting state, (b) focused-attention meditation (FAM) and (c) open-monitoringmeditation (OMM) in a group of expert meditators (12 monks).To this end, weestimated MEG source time courses using a minimum-norm solution and computed (1)spectral power in multiple frequency bands (delta, theta, alpha, beta andgamma), (2) graph theoretical measures, (3) long-range coupling using imaginarycoherence and weighed phase-lag index and (4) multifractal scaling parameters using Wavelet Leader-based Multifractal formalism. We compared all the measures in the three conditions(OMM, FAM and resting state) and tested for statistical significance using permutationtest (paired t-test) corrected by maximum statistics. We also used a machinelearning framework in order to see which features provide the highestclassification across conditions. Our findings reveal several differencesbetween FAM, OMM and the resting-state condition. Compared to OMM, FAM isassociated with an increase in power in regions involved in attention andperformance monitoring. In OMM, increases in activity were observed in regionsinvolved in memory and emotion processing. Moreover, OMM seems to have strongestand more connections, while resting state have connections that are weaker andfewer in number compared to OMM and FAM. We discuss these results in thecontext of previous cognitive neuroimaging studies of meditation and paths forfuture research are proposed.

The phenomenology and reported effects of meditation vary according to the technique practiced. While numerous studies have explored the cerebral mechanisms involved in meditation, little research provides direct comparisons between the neuronal network dynamics involved in different meditation techniques. Here, we explore and compare brain signals recorded with magnetoencephalography (MEG) during (a) focused-attention meditation (FAM), and (b) open-monitoring meditation (OMM) in a group of expert meditators (12 monks). To this end, we estimated MEG source time courses using minimum-norm and computed spectral power in multiple frequency bands (delta, theta, alpha, beta and gamma), graph theoretical measures and multifractal scaling parameters in both conditions. Preliminary findings reveal several differences between FAM and OMM. Interestingly, OMM was associated with higher theta power in the right temporal pole. We discuss these results in the context of previous cognitive neuroimaging studies of meditation and paths for future research are proposed.

Background. Electrocorticography (ECoG) measures the distribution of electrical potentials by means of electrodes grids implanted close to the cortical surface. A full interpretation of ECoG data requires solving the ill-posed inverse problem of reconstructing the spatio-temporal distribution of neural currents responsible for the recorded signals. Only in the last few years some methods have been proposed to solve this inverse problem [1]. Methods. This study [2] addresses the ECoG source modelling using a beamformer method. We computed the lead-field matrix which maps the neural currents onto the sensors space by a novel routine provided by the OpenMEEG framework [3]. The ECoG source-modeling problem requires to invert this matrix by means of a regularization method which reduces its intrinsic numerical instability: we performed an analysis of the condition number of the lead-field matrix for different configurations of the electrodes grid. Finally, we provided quantitative results for source modeling using a Linear Constraint Minimum Variance (LCMV) beamformer [4]. The validation of the effectiveness of beamforming in ECoG was performed both with synthetic data and with experimental data recorded during a rapid visual categorization task. Results. For all considered grids the condition number indicates that the ECoG inverse problem is mildly ill-conditioned. For realistic SNR we found a good performance of the LCMV algorithm for both localization and waveforms reconstruction. The flow of information reconstructed by analyzing real data seems consistent with both invasive monkey electrophysiology studies and non-invasive (MEG and fMRI) human studies. References: 1. Dumpelmann et al., (2012), Human brain mapping, 33(5), 1172-1188 2. Pascarella et al. (2016), Journal of Neuroscience Methods, 263(5), 134-144 3. Kybic et al., (2005), Medical Imaging, IEEE Transactions on, 24(1), 12-28 4. Van Veen et al., (1997), Biomedical Engineering, IEEE Transactions on, 44(9), 867-880