Automatic bad channel detection in intracranial electroencephalographic recordings using ensemble machine learning

SEEG electrode shaft affects amplitude and latency of potentials evoked with single pulse electrical stimulation

BACKGROUND

Long and thin shaft electrodes are implanted intracerebrally for stereoelectroencephalography (SEEG) in patients with pharmacoresistant focal epilepsies. Two adjacent contacts of one of such electrodes can deliver a train of single pulse electrical stimulations (SPES), and evoked potentials (EPs) are recorded on other contacts. In this study we assess if stimulating and recording on the same shaft, as opposed to different shafts, has an impact on common EP features.

NEW METHOD

We leverage the large volume of SEEG data gathered in the F-TRACT database and analyze data from nearly one thousand SEEG implantations in order to verify whether stimulation and recording from the same shaft influence the EP pattern.

RESULTS

We found that when the stimulated and the recording contacts were located on the same shaft, the mean and median amplitudes of an EP are greater, and its mean and median latencies are smaller than when the contacts were located on different shafts. This effect is small (Cohen's d ∼ 0.1), but robust (p-value < 10-3) across the SEEG database.

COMPARISON WITH EXISTING METHOD(S)

Our study is the first one to address this question. Due to the choice of commonly used EP features, our method is congruent with other studies.

CONCLUSIONS

The magnitude of the reported effect does not obligate all standard analyses to correct for it, unless they aim at high precision. The source of the effect is not clear. Manufacturers of SEEG electrodes could examine it and potentially minimize the effect in their future products.

Benchmarking signal quality and spatiotemporal distribution of interictal spikes in prolonged human iEEG recordings using CorTec wireless brain interchange

Neuromodulation through implantable pulse generators (IPGs) represents an important treatment approach for neurological disorders. While the field has observed the success of state-of-the-art interventions, such as deep brain stimulation (DBS) or responsive neurostimulation (RNS), implantable systems face various technical challenges, including the restriction of recording from a limited number of brain sites, power management, and limited external access to the assessed neural data in a continuous fashion. To the best of our knowledge, for the first time in this study, we investigated the feasibility of recording human intracranial EEG (iEEG) using a benchtop version of the Brain Interchange (BIC) unit of CorTec, which is a portable, wireless, and externally powered implant with sensing and stimulation capabilities. We developed a MATLAB/SIMULINK-based rapid prototyping environment and a graphical user interface (GUI) to acquire and visualize the iEEG captured from all 32 channels of the BIC unit. We recorded prolonged iEEG (~ 24 h) from three human subjects with externalized depth leads using the BIC and commercially available clinical amplifiers simultaneously in the epilepsy monitoring unit (EMU). The iEEG signal quality of both streams was compared, and the results demonstrated a comparable power spectral density (PSD) in all the systems in the low-frequency band (< 80 Hz). However, notable differences were primarily observed above 100 Hz, where the clinical amplifiers were associated with lower noise floor (BIC-17 dB vs. clinical amplifiers <  - 25 dB). We employed an established spike detector to assess and compare the spike rates in each iEEG stream. We observed over 90% conformity between the spikes rates and their spatial distribution captured with BIC and clinical systems. Additionally, we quantified the packet loss characteristic in the iEEG signal during the wireless data transfer and conducted a series of simulations to compare the performance of different interpolation methods for recovering the missing packets in signals at different frequency bands. We noted that simple linear interpolation has the potential to recover the signal and reduce the noise floor with modest packet loss levels reaching up to 10%. Overall, our results indicate that while tethered clinical amplifiers exhibited noticeably better noise floor above 80 Hz, epileptic spikes can still be detected successfully in the iEEG recorded with the externally powered wireless BIC unit opening the road for future closed-loop neuromodulation applications with continuous access to brain activity.

Lateralization and Time-Course of Cortical Phonological Representations during Syllable Production

Spoken language contains information at a broad range of timescales, from phonetic distinctions on the order of milliseconds to semantic contexts which shift over seconds to minutes. It is not well understood how the brain's speech production systems combine features at these timescales into a coherent vocal output. We investigated the spatial and temporal representations in cerebral cortex of three phonological units with different durations: consonants, vowels, and syllables. Electrocorticography (ECoG) recordings were obtained from five participants while speaking single syllables. We developed a novel clustering and Kalman filter-based trend analysis procedure to sort electrodes into temporal response profiles. A linear discriminant classifier was used to determine how strongly each electrode's response encoded phonological features. We found distinct time-courses of encoding phonological units depending on their duration: consonants were represented more during speech preparation, vowels were represented evenly throughout trials, and syllables during production. Locations of strongly speech-encoding electrodes (the top 30% of electrodes) likewise depended on phonological element duration, with consonant-encoding electrodes left-lateralized, vowel-encoding hemispherically balanced, and syllable-encoding right-lateralized. The lateralization of speech-encoding electrodes depended on onset time, with electrodes active before or after speech production favoring left hemisphere and those active during speech favoring the right. Single-electrode speech classification revealed cortical areas with preferential encoding of particular phonemic elements, including consonant encoding in the left precentral and postcentral gyri and syllable encoding in the right middle frontal gyrus. Our findings support neurolinguistic theories of left hemisphere specialization for processing short-timescale linguistic units and right hemisphere processing of longer-duration units.

Communication dynamics in the human connectome shape the cortex-wide propagation of direct electrical stimulation

Communication between gray matter regions underpins all facets of brain function. We study inter-areal communication in the human brain using intracranial EEG recordings, acquired following 29,055 single-pulse direct electrical stimulations in a total of 550 individuals across 20 medical centers (average of 87 ± 37 electrode contacts per subject). We found that network communication models-computed on structural connectivity inferred from diffusion MRI-can explain the causal propagation of focal stimuli, measured at millisecond timescales. Building on this finding, we show that a parsimonious statistical model comprising structural, functional, and spatial factors can accurately and robustly predict cortex-wide effects of brain stimulation (R²=46% in data from held-out medical centers). Our work contributes toward the biological validation of concepts in network neuroscience and provides insight into how connectome topology shapes polysynaptic inter-areal signaling. We anticipate that our findings will have implications for research on neural communication and the design of brain stimulation paradigms.

Propofol anesthesia-induced spatiotemporal changes in cortical activity with loss of external and internal awareness: An electrocorticography study

OBJECTIVE

To understand the underlying mechanism of consciousness, investigating spatiotemporal changes in the cortical activity during the induction phase of unconsciousness is important. Loss of consciousness induced by general anesthesia is not necessarily accompanied by a uniform inhibition of all cortical activities. We hypothesized that cortical regions involved in internal awareness would be suppressed after disruption of cortical regions involved in external awareness. Thus, we investigated temporal changes in cortex during induction of unconsciousness.

METHODS

We recorded electrocorticography data of 16 epilepsy patients and investigated power spectral changes during induction phase from awake state to unconsciousness. Temporal changes were assessed at 1) the start point and 2) the interval of normalized time between start and end of power change (Δ t^normalized).

RESULTS

We found that the power increased at frequencies < 46 Hz, and decreased in range of 62-150 Hz, in global channels. In temporal changes of power change, superior parietal lobule and dorsolateral prefrontal cortex started to change early, but the changes were completed over a prolonged interval, whereas angular gyrus and associative visual cortex showed a delayed change and rapid completion.

CONCLUSIONS

Loss of consciousness induced by general anesthesia results first from disrupted communication between self and external world, followed by disrupted communication within self, with decreased activities of superior parietal lobule and dorsolateral prefrontal cortex, and later, attenuated activities of angular gyrus.

SIGNIFICANCE

Our findings provided neurophysiological evidence for the temporal changes in consciousness components induced by general anesthesia.

Relevance-based channel selection in motor imagery brain-computer interface

Objective.Channel selection in the electroencephalogram (EEG)-based brain-computer interface (BCI) has been extensively studied for over two decades, with the goal being to select optimal subject-specific channels that can enhance the overall decoding efficacy of the BCI. With the emergence of deep learning (DL)-based BCI models, there arises a need for fresh perspectives and novel techniques to conduct channel selection. In this regard, subject-independent channel selection is relevant, since DL models trained using cross-subject data offer superior performance, and the impact of inherent inter-subject variability of EEG characteristics on subject-independent DL training is not yet fully understood.Approach.Here, we propose a novel methodology for implementing subject-independent channel selection in DL-based motor imagery (MI)-BCI, using layer-wise relevance propagation (LRP) and neural network pruning. Experiments were conducted using Deep ConvNet and 62-channel MI data from the Korea University EEG dataset.Main Results.Using our proposed methodology, we achieved a 61% reduction in the number of channels without any significant drop (p = 0.09) in subject-independent classification accuracy, due to the selection of highly relevant channels by LRP. LRP relevance-based channel selections provide significantly better accuracies compared to conventional weight-based selections while using less than 40% of the total number of channels, with differences in accuracies ranging from 5.96% to 1.72%. The performance of the adapted sparse-LRP model using only 16% of the total number of channels is similar to that of the adapted baseline model (p = 0.13). Furthermore, the accuracy of the adapted sparse-LRP model using only 35% of the total number of channels exceeded that of the adapted baseline model by 0.53% (p = 0.81). Analyses of channels chosen by LRP confirm the neurophysiological plausibility of selection, and emphasize the influence of motor, parietal, and occipital channels in MI-EEG classification.Significance.The proposed method addresses a traditional issue in EEG-BCI decoding, while being relevant and applicable to the latest developments in the field of BCI. We believe that our work brings forth an interesting and important application of model interpretability as a problem-solving technique.

ABOT: an open-source online benchmarking tool for machine learning-based artefact detection and removal methods from neuronal signals

Brain signals are recorded using different techniques to aid an accurate understanding of brain function and to treat its disorders. Untargeted internal and external sources contaminate the acquired signals during the recording process. Often termed as artefacts, these contaminations cause serious hindrances in decoding the recorded signals; hence, they must be removed to facilitate unbiased decision-making for a given investigation. Due to the complex and elusive manifestation of artefacts in neuronal signals, computational techniques serve as powerful tools for their detection and removal. Machine learning (ML) based methods have been successfully applied in this task. Due to ML’s popularity, many articles are published every year, making it challenging to find, compare and select the most appropriate method for a given experiment. To this end, this paper presents ABOT (Artefact removal Benchmarking Online Tool) as an online benchmarking tool which allows users to compare existing ML-driven artefact detection and removal methods from the literature. The characteristics and related information about the existing methods have been compiled as a knowledgebase (KB) and presented through a user-friendly interface with interactive plots and tables for users to search it using several criteria. Key characteristics extracted from over 120 articles from the literature have been used in the KB to help compare the specific ML models. To comply with the FAIR (Findable, Accessible, Interoperable and Reusable) principle, the source code and documentation of the toolbox have been made available via an open-access repository.

A sparse representation strategy to eliminate pseudo-HFO events from intracranial EEG for seizure onset zone localization

Objective.High-frequency oscillations (HFOs) are considered a biomarker of the epileptogenic zone in intracranial EEG recordings. However, automated HFO detectors confound true oscillations with spurious events caused by the presence of artifacts.Approach.We hypothesized that, unlike pseudo-HFOs with sharp transients or arbitrary shapes, real HFOs have a signal characteristic that can be represented using a small number of oscillatory bases. Based on this hypothesis using a sparse representation framework, this study introduces a new classification approach to distinguish true HFOs from the pseudo-events that mislead seizure onset zone (SOZ) localization. Moreover, we further classified the HFOs into ripples and fast ripples by introducing an adaptive reconstruction scheme using sparse representation. By visualizing the raw waveforms and time-frequency representation of events recorded from 16 patients, three experts labeled 6400 candidate events that passed an initial amplitude-threshold-based HFO detector. We formed a redundant analytical multiscale dictionary built from smooth oscillatory Gabor atoms and represented each event with orthogonal matching pursuit by using a small number of dictionary elements. We used the approximation error and residual signal at each iteration to extract features that can distinguish the HFOs from any type of artifact regardless of their corresponding source. We validated our model on sixteen subjects with thirty minutes of continuous interictal intracranial EEG recording from each.Main results.We showed that the accuracy of SOZ detection after applying our method was significantly improved. In particular, we achieved a 96.65% classification accuracy in labeled events and a 17.57% improvement in SOZ detection on continuous data. Our sparse representation framework can also distinguish between ripples and fast ripples.Significance.We show that by using a sparse representation approach we can remove the pseudo-HFOs from the pool of events and improve the reliability of detected HFOs in large data sets and minimize manual artifact elimination.

Brain mapping of auditory hallucinations and illusions induced by direct intracortical electrical stimulation

BACKGROUND

The exact architecture of the human auditory cortex remains a subject of debate, with discrepancies between functional and microstructural studies. In a hierarchical framework for sensory perception, simple sound perception is expected to take place in the primary auditory cortex, while the processing of complex, or more integrated perceptions is proposed to rely on associative and higher-order cortices.

OBJECTIVES

We hypothesize that auditory symptoms induced by direct electrical stimulation (DES) offer a window into the architecture of the brain networks involved in auditory hallucinations and illusions. The intracranial recordings of these evoked perceptions of varying levels of integration provide the evidence to discuss the theoretical model.

METHODS

We analyzed SEEG recordings from 50 epileptic patients presenting auditory symptoms induced by DES. First, using the Juelich cytoarchitectonic parcellation, we quantified which regions induced auditory symptoms when stimulated (ROI approach). Then, for each evoked auditory symptom type (illusion or hallucination), we mapped the cortical networks showing concurrent high-frequency activity modulation (HFA approach).

RESULTS

Although on average, illusions were found more laterally and hallucinations more posteromedially in the temporal lobe, both perceptions were elicited in all levels of the sensory hierarchy, with mixed responses found in the overlap. The spatial range was larger for illusions, both in the ROI and HFA approaches. The limbic system was specific to the hallucinations network, and the inferior parietal lobule was specific to the illusions network.

DISCUSSION

Our results confirm a network-based organization underlying conscious sound perception, for both simple and complex components. While symptom localization is interesting from an epilepsy semiology perspective, the hallucination-specific modulation of the limbic system is particularly relevant to tinnitus and schizophrenia.

Intracerebral mechanisms explaining the impact of incidental feedback on mood state and risky choice

Identifying factors whose fluctuations are associated with choice inconsistency is a major issue for rational decision theory. Here, we investigated the neuro-computational mechanisms through which mood fluctuations may bias human choice behavior. Intracerebral EEG data were collected in a large group of subjects (n=30) while they were performing interleaved quiz and choice tasks that were designed to examine how a series of unrelated feedbacks affect decisions between safe and risky options. Neural baseline activity preceding choice onset was confronted first to mood level, estimated by a computational model integrating the feedbacks received in the quiz task, and then to the weighting of option attributes, in a computational model predicting risk attitude in the choice task. Results showed that (1) elevated broadband gamma activity (BGA) in the ventromedial prefrontal cortex (vmPFC) and dorsal anterior insula (daIns) was respectively signaling periods of high and low mood, (2) increased vmPFC and daIns BGA respectively promoted and tempered risk taking by overweighting gain vs. loss prospects. Thus, incidental feedbacks induce brain states that correspond to different moods and bias the evaluation of risky options. More generally, these findings might explain why people experiencing positive (or negative) outcome in some part of their life tend to expect success (or failure) in any other.

Advances in human intracranial electroencephalography research, guidelines and good practices

Since the second-half of the twentieth century, intracranial electroencephalography (iEEG), including both electrocorticography (ECoG) and stereo-electroencephalography (sEEG), has provided an intimate view into the human brain. At the interface between fundamental research and the clinic, iEEG provides both high temporal resolution and high spatial specificity but comes with constraints, such as the individual's tailored sparsity of electrode sampling. Over the years, researchers in neuroscience developed their practices to make the most of the iEEG approach. Here we offer a critical review of iEEG research practices in a didactic framework for newcomers, as well addressing issues encountered by proficient researchers. The scope is threefold: (i) review common practices in iEEG research, (ii) suggest potential guidelines for working with iEEG data and answer frequently asked questions based on the most widespread practices, and (iii) based on current neurophysiological knowledge and methodologies, pave the way to good practice standards in iEEG research. The organization of this paper follows the steps of iEEG data processing. The first section contextualizes iEEG data collection. The second section focuses on localization of intracranial electrodes. The third section highlights the main pre-processing steps. The fourth section presents iEEG signal analysis methods. The fifth section discusses statistical approaches. The sixth section draws some unique perspectives on iEEG research. Finally, to ensure a consistent nomenclature throughout the manuscript and to align with other guidelines, e.g., Brain Imaging Data Structure (BIDS) and the OHBM Committee on Best Practices in Data Analysis and Sharing (COBIDAS), we provide a glossary to disambiguate terms related to iEEG research.

Decoding Intracranial EEG With Machine Learning: A Systematic Review

Advances in intracranial electroencephalography (iEEG) and neurophysiology have enabled the study of previously inaccessible brain regions with high fidelity temporal and spatial resolution. Studies of iEEG have revealed a rich neural code subserving healthy brain function and which fails in disease states. Machine learning (ML), a form of artificial intelligence, is a modern tool that may be able to better decode complex neural signals and enhance interpretation of these data. To date, a number of publications have applied ML to iEEG, but clinician awareness of these techniques and their relevance to neurosurgery, has been limited. The present work presents a review of existing applications of ML techniques in iEEG data, discusses the relative merits and limitations of the various approaches, and examines potential avenues for clinical translation in neurosurgery. One-hundred-seven articles examining artificial intelligence applications to iEEG were identified from 3 databases. Clinical applications of ML from these articles were categorized into 4 domains: i) seizure analysis, ii) motor tasks, iii) cognitive assessment, and iv) sleep staging. The review revealed that supervised algorithms were most commonly used across studies and often leveraged publicly available timeseries datasets. We conclude with recommendations for future work and potential clinical applications.

GATSMOTE: Improving Imbalanced Node Classification on Graphs via Attention and Homophily

In recent decades, non-invasive neuroimaging techniques and graph theories have enabled a better understanding of the structural patterns of the human brain at a macroscopic level. As one of the most widely used non-invasive techniques, an electroencephalogram (EEG) may collect non-neuronal signals from “bad channels”. Automatically detecting these bad channels represents an imbalanced classification task; research on the topic is rather limited. Because the human brain can be naturally modeled as a complex graph network based on its structural and functional characteristics, we seek to extend previous imbalanced node classification techniques to the bad-channel detection task. We specifically propose a novel edge generator considering the prominent small-world organization of the human brain network. We leverage the attention mechanism to adaptively calculate the weighted edge connections between each node and its neighboring nodes. Moreover, we follow the homophily assumption in graph theory to add edges between similar nodes. Adding new edges between nodes sharing identical labels shortens the path length, thus facilitating low-cost information messaging.

Open multimodal iEEG-fMRI dataset from naturalistic stimulation with a short audiovisual film

Intracranial human recordings are a valuable and rare resource of information about the brain. Making such data publicly available not only helps tackle reproducibility issues in science, it helps make more use of these valuable data. This is especially true for data collected using naturalistic tasks. Here, we describe a dataset collected from a large group of human subjects while they watched a short audiovisual film. The dataset has several unique features. First, it includes a large amount of intracranial electroencephalography (iEEG) data (51 participants, age range of 5-55 years, who all performed the same task). Second, it includes functional magnetic resonance imaging (fMRI) recordings (30 participants, age range of 7-47) during the same task. Eighteen participants performed both iEEG and fMRI versions of the task, non-simultaneously. Third, the data were acquired using a rich audiovisual stimulus, for which we provide detailed speech and video annotations. This dataset can be used to study neural mechanisms of multimodal perception and language comprehension, and similarity of neural signals across brain recording modalities.

A brain atlas of axonal and synaptic delays based on modelling of cortico-cortical evoked potentials

Epilepsy presurgical investigation may include focal intracortical single-pulse electrical stimulations with depth electrodes, which induce cortico-cortical evoked potentials at distant sites because of white matter connectivity. Cortico-cortical evoked potentials provide a unique window on functional brain networks because they contain sufficient information to infer dynamical properties of large-scale brain connectivity, such as preferred directionality and propagation latencies.

 Here, we developed a biologically informed modelling approach to estimate the neural physiological parameters of brain functional networks from the cortico-cortical evoked potentials recorded in a large multicentric database. Specifically, we considered each cortico-cortical evoked potential as the output of a transient stimulus entering the stimulated region, which directly propagated to the recording region. Both regions were modelled as coupled neural mass models, the parameters of which were estimated from the first cortico-cortical evoked potential component, occurring before 80 ms, using dynamic causal modelling and Bayesian model inversion. This methodology was applied to the data of 780 patients with epilepsy from the F-TRACT database, providing a total of 34 354 bipolar stimulations and 774 445 cortico-cortical evoked potentials. The cortical mapping of the local excitatory and inhibitory synaptic time constants and of the axonal conduction delays between cortical regions was obtained at the population level using anatomy-based averaging procedures, based on the Lausanne2008 and the HCP-MMP1 parcellation schemes, containing 130 and 360 parcels, respectively. To rule out brain maturation effects, a separate analysis was performed for older (>15 years) and younger patients (<15 years). In the group of older subjects, we found that the cortico-cortical axonal conduction delays between parcels were globally short (median = 10.2 ms) and only 16% were larger than 20 ms. This was associated to a median velocity of 3.9 m/s. Although a general lengthening of these delays with the distance between the stimulating and recording contacts was observed across the cortex, some regions were less affected by this rule, such as the insula for which almost all efferent and afferent connections were faster than 10 ms. Synaptic time constants were found to be shorter in the sensorimotor, medial occipital and latero-temporal regions, than in other cortical areas. Finally, we found that axonal conduction delays were significantly larger in the group of subjects younger than 15 years, which corroborates that brain maturation increases the speed of brain dynamics.

To our knowledge, this study is the first to provide a local estimation of axonal conduction delays and synaptic time constants across the whole human cortex in vivo, based on intracerebral electrophysiological recordings.

Investigation on Robustness of EEG-based Brain-Computer Interfaces

Electroencephalogram (EEG)-based brain-computer interface (BCI) systems tend to suffer from performance degradation due to the presence of noise and artifacts in EEG data. This study is aimed at systematically investigating the robustness of state-of-the-art machine learning and deep learning based EEG-BCI models for motor imagery classification against simulated channel-specific noise in EEG data, at various low values of signal-to-noise ratio (SNR). Our results illustrate higher robustness of deep learning based MI classification models compared to the traditional machine learning based model, while identifying a set of channels with large sensitivity to simulated channel-specific noise. The EEGNet is relatively more robust towards channel-specific noise than Shallow ConvNet and FBCSP. We propose a preliminary solution, based on activation function, to improve the robustness of the deep learning models. By using saturating nonlinearities, the percentage drop in classification accuracy for SNR of -18 dB had reduced from 10.99% to 6.53% for EEGNet and 14.05% to 3.57% for Shallow ConvNet. Through this study, we emphasize the need for a more precise solution for enhancing the robustness, and thereby usability of EEG-BCI systems.

Anatomical dissociation of intracerebral signals for reward and punishment prediction errors in humans

Whether maximizing rewards and minimizing punishments rely on distinct brain systems remains debated, given inconsistent results coming from human neuroimaging and animal electrophysiology studies. Bridging the gap across techniques, we recorded intracerebral activity from twenty participants while they performed an instrumental learning task. We found that both reward and punishment prediction errors (PE), estimated from computational modeling of choice behavior, correlate positively with broadband gamma activity (BGA) in several brain regions. In all cases, BGA scaled positively with the outcome (reward or punishment versus nothing) and negatively with the expectation (predictability of reward or punishment). However, reward PE were better signaled in some regions (such as the ventromedial prefrontal and lateral orbitofrontal cortex), and punishment PE in other regions (such as the anterior insula and dorsolateral prefrontal cortex). These regions might therefore belong to brain systems that differentially contribute to the repetition of rewarded choices and the avoidance of punished choices.

Whether maximizing rewards and minimizing punishments rely on distinct brain learning systems remains debated. Here, using intracerebral recordings in humans, the authors provide evidence for brain regions differentially engaged in signaling reward and punishment prediction errors that prescribe repetition versus avoidance of past choices.

A Review on Machine Learning for EEG Signal Processing in Bioengineering

Electroencephalography (EEG) has been a staple method for identifying certain health conditions in patients since its discovery. Due to the many different types of classifiers available to use, the analysis methods are also equally numerous. In this review, we will be examining specifically machine learning methods that have been developed for EEG analysis with bioengineering applications. We reviewed literature from 1988 to 2018 to capture previous and current classification methods for EEG in multiple applications. From this information, we are able to determine the overall effectiveness of each machine learning method as well as the key characteristics. We have found that all the primary methods used in machine learning have been applied in some form in EEG classification. This ranges from Naive-Bayes to Decision Tree/Random Forest, to Support Vector Machine (SVM). Supervised learning methods are on average of higher accuracy than their unsupervised counterparts. This includes SVM and KNN. While each of the methods individually is limited in their accuracy in their respective applications, there is hope that the combination of methods when implemented properly has a higher overall classification accuracy. This paper provides a comprehensive overview of Machine Learning applications used in EEG analysis. It also gives an overview of each of the methods and general applications that each is best suited to.

Probabilistic functional tractography of the human cortex revisited

In patients with pharmaco-resistant focal epilepsies investigated with intracranial electroencephalography (iEEG), direct electrical stimulations of a cortical region induce cortico-cortical evoked potentials (CCEP) in distant cerebral cortex, which properties can be used to infer large scale brain connectivity. In 2013, we proposed a new probabilistic functional tractography methodology to study human brain connectivity. We have now been revisiting this method in the F-TRACT project (f-tract.eu) by developing a large multicenter CCEP database of several thousand stimulation runs performed in several hundred patients, and associated processing tools to create a probabilistic atlas of human cortico-cortical connections. Here, we wish to present a snapshot of the methods and data of F-TRACT using a pool of 213 epilepsy patients, all studied by stereo-encephalography with intracerebral depth electrodes. The CCEPs were processed using an automated pipeline with the following consecutive steps: detection of each stimulation run from stimulation artifacts in raw intracranial EEG (iEEG) files, bad channels detection with a machine learning approach, model-based stimulation artifact correction, robust averaging over stimulation pulses. Effective connectivity between the stimulated and recording areas is then inferred from the properties of the first CCEP component, i.e. onset and peak latency, amplitude, duration and integral of the significant part. Finally, group statistics of CCEP features are implemented for each brain parcel explored by iEEG electrodes. The localization (coordinates, white/gray matter relative positioning) of electrode contacts were obtained from imaging data (anatomical MRI or CT scans before and after electrodes implantation). The iEEG contacts were repositioned in different brain parcellations from the segmentation of patients' anatomical MRI or from templates in the MNI coordinate system. The F-TRACT database using the first pool of 213 patients provided connectivity probability values for 95% of possible intrahemispheric and 56% of interhemispheric connections and CCEP features for 78% of intrahemisheric and 14% of interhemispheric connections. In this report, we show some examples of anatomo-functional connectivity matrices, and associated directional maps. We also indicate how CCEP features, especially latencies, are related to spatial distances, and allow estimating the velocity distribution of neuronal signals at a large scale. Finally, we describe the impact on the estimated connectivity of the stimulation charge and of the contact localization according to the white or gray matter. The most relevant maps for the scientific community are available for download on f-tract. eu (David et al., 2017) and will be regularly updated during the following months with the addition of more data in the F-TRACT database. This will provide an unprecedented knowledge on the dynamical properties of large fiber tracts in human.