Search

Magnetoencephalography and electroencephalography (M/EEG) seed-based connectivity analysis typically requires regions of interest (ROI)-based extraction of measures. M/EEG ROI-derived source activity can be treated in different ways. For instance, it is possible to average each ROI's time series prior to calculating connectivity measures. Alternatively one can compute connectivity maps for each element of the ROI, prior to dimensionality reduction to obtain a single map. The impact of these different strategies on connectivity estimation is still unclear.Here, we address this question within a large MEG resting state cohort (N=113) and simulated data. We consider 68 ROIs (Desikan-Kiliany atlas), two measures of connectivity (phase locking value-PLV, and its imaginary counterpart- ciPLV), and three frequency bands (theta 4-8 Hz, alpha 9-12 Hz, beta 15-30 Hz). We consider four extraction methods: (i) mean, or (ii) PCA of the activity within the ROI before computing connectivity, (iii) average, or (iv) maximum connectivity after computing connectivity for each element of the seed. Connectivity outputs from these extraction strategies are then compared with hierarchical clustering, followed by direct contrasts across extraction methods. Finally, the results are validated by using a set of realistic simulations.We show that ROI-based connectivity maps vary remarkably across strategies in both connectivity magnitude and spatial distribution. Dimensionality reduction procedures conducted after computing connectivity are more similar to each-other, while PCA before approach is the most dissimilar to other approaches. Although differences across methods are consistent across frequency bands, they are influenced by the connectivity metric and ROI size. Greater differences were observed for ciPLV than PLV, and in larger ROIs. Realistic simulations confirmed that after aggregation procedures are generally more accurate but have lower specificity (higher rate of false positive connections). Although computationally demanding, after dimensionality reduction strategies should be preferred when higher sensitivity is desired. Given the remarkable differences across aggregation procedures, caution is warranted in comparing results across studies applying different extraction methods.

The accurate characterization of cortical functional connectivity from Magnetoencephalography (MEG) data remains a challenging problem due to the subjective nature of the analysis, which requires several decisions at each step of the analysis pipeline, such as the choice of a source estimation algorithm, a connectivity metric and a cortical parcellation, to name but a few. Recent studies have emphasized the importance of selecting the regularization parameter in minimum norm estimates with caution, as variations in its value can result in significant differences in connectivity estimates. In particular, the amount of regularization that is optimal for MEG source estimation can actually be suboptimal for coherence-based MEG connectivity analysis. In this study, we expand upon previous work by examining a broader range of commonly used connectivity metrics, including the imaginary part of coherence, corrected imaginary part of Phase Locking Value, and weighted Phase Lag Index, within a larger and more realistic simulation scenario. Our results show that the best estimate of connectivity is achieved using a regularization parameter that is 1 or 2 orders of magnitude smaller than the one that yields the best source estimation. This remarkable difference may imply that previous work assessing source-space connectivity using minimum-norm may have benefited from using less regularization, as this may have helped reduce false positives. Importantly, we provide the code for MEG data simulation and analysis, offering the research community a valuable open source tool for informed selections of the regularization parameter when using minimum-norm for source space connectivity analyses.

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), magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI). 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.

Electrophysiological source imaging (ESI) aims at reconstructing the precise origin of brain activity from measurements of the electric field on the scalp. Across laboratories/research centers/hospitals, ESI is performed with different methods, partly due to the ill-posedness of the underlying mathematical problem. However, it is difficult to find systematic comparisons involving a wide variety of methods. Further, existing comparisons rarely take into account the variability of the results with respect to the input parameters. Finally, comparisons are typically performed using either synthetic data, or in-vivo data where the ground-truth is only roughly known. 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 precisely known. We compare ten different ESI methods, using their implementation in the MNE-Python package: MNE, dSPM, LORETA, sLORETA, eLORETA, LCMV beamformers, irMxNE, Gamma Map, SESAME and dipole fitting. We perform comparisons under multiple choices of input parameters, to assess the accuracy of the best reconstruction, as well as the impact of such parameters on the localization performance. Best reconstructions often fall within 1 cm from the true source, with most accurate methods hitting an average localization error of 1.2 cm and outperforming least accurate ones erring by 2.5 cm. As expected, dipolar and sparsity-promoting methods tend to outperform distributed methods. For several distributed methods, the best regularization parameter turned out to be the one in principle associated with low SNR, despite the high SNR of the available dataset. Depth weighting played no role for two out of the six methods implementing it. Sensitivity to input parameters varied widely between methods. While one would expect high variability being associated with low localization error at the best solution, this is not always the case, with some methods producing highly variable results and high localization error, and other methods producing stable results with low localization error. In particular, recent dipolar and sparsity-promoting methods provide significantly better results than older distributed methods. As we repeated the tests with “conventional” (32 channels) and dense (64, 128, 256 channels) EEG recordings, we observed little impact of the number of channels on localization accuracy; however, for distributed methods denser montages provide smaller spatial dispersion. Overall findings confirm that EEG is a reliable technique for localization of point sources and therefore 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.

Neural oscillations contribute to speech parsing via cortical tracking of hierarchical linguistic structures, including syllable rate. While the properties of neural entrainment have been largely probed with speech stimuli at either normal or artificially accelerated rates, the important case of natural fast speech has been largely overlooked. Using magnetoencephalography, we found that listening to naturally-produced speech was associated with cortico-acoustic coupling, both at normal (∼6 syllables/s) and fast (∼9 syllables/s) rates, with a corresponding shift in peak entrainment frequency. Interestingly, time-compressed sentences did not yield such coupling, despite being generated at the same rate as the natural fast sentences. Additionally, neural activity in right motor cortex exhibited stronger tuning to natural fast rather than to artificially accelerated speech, and showed evidence for stronger phase-coupling with left temporo-parietal and motor areas. These findings are highly relevant for our understanding of the role played by auditory and motor cortex oscillations in the perception of naturally produced speech.

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.