Background: Biological age is a key concept in the development of biomarkers of health and disease. We develop a prediction of the functional autonomic age (FAA) from infancy to adolescence based on the ECG-derived tachogram recorded at the onset of N2 sleep. Methods: A cohort of ECG recordings from 1004 typically developing infants, children and adolescents (age range: 1 month to 17 years) was used to train feature-based and deep neural network-based regression models for the prediction of FAA. Weighted mean absolute error (wMAE) was used to define accuracy and evaluated with 10-fold cross-validation. Effect size was used to compare model accuracies and linear regression was used to evaluate confounds. The combination of FAA with an EEG-based estimate of functional brain age (FBA) was also tested. Results: A feature-based FAA had a wMAE of 1.78 years (95 %CI: 1.62-1.93, n = 1004) and was comparable to deep neural network regression (wMAE = 1.85 years, 95 %CI: 1.66-1.98). Accuracy was affected by age and age2 (t = -4.97, p < 0.001 and t = 9.66, p < 0.001, respectively) with smaller errors at younger ages, but not biological sex (t = -0.660, p = 0.510). Combining the FAA with a functional brain age derived from the EEG resulted improved accuracy, with neural networks-based methods superior (wMAE of 0.81 years, 95 %CI: 0.73-0.88, effect size D = 0.77, 95 %CI: 0.70-0.84, n = 1004). Conclusion: FAA derived from the tachogram accurately represents age from infancy to adolescence. The combination of FAA with FBA improves age prediction accuracy.

Stereo-EEG (SEEG) cortical stimulation enables individualized mapping of language networks. Regions associated with induced language deficits are marked as 'language-positive' and considered important in supporting function. It remains unclear, however, whether small lesions in 'language-positive' sites created by radiofrequency thermocoagulation are sufficient to cause language deficits. Thirty-six consecutive SEEG patients with drug-resistant focal epilepsy were prospectively recruited from two Australian epilepsy centres. Formal language assessment was undertaken before and 3 months after radiofrequency thermocoagulation [mean = 106.92 days, standard deviation (SD) = 27.83], which included the Boston Naming Test, Auditory Naming Test and semantic fluency task. During high-frequency (50 Hz) cortical stimulation, language was assessed in vivo using visual and auditory naming, reading, spontaneous speech and/or counting tasks. To evaluate group changes post radiofrequency thermocoagulation, paired sample t-tests were undertaken. Reliable change indices were calculated to classify language decline, and independent samples t-tests or χ2 tests were then used to compare groups on selected clinical and demographic variables. Of the 36 patients (mean = 36.19 years old, SD = 9.22 years, range = 17-56 years, 56% female), 14 (39%) had a language-dominant epileptogenic zone (EZ), 18 (50%) a non-dominant EZ and 4 (11%) a bilateral EZ. A mean of 12.28 (SD = 6.84, range = 2-29) coagulation sites were undertaken per patient. Language decline was associated with radiofrequency thermocoagulation of a language-positive site [χ12 = 6.94, P = 0.008, moderate effect size odds ratio = 10.00, 95% confidence interval (1.68, 59.31)]; specifically, 63% (5/8) of patients with radiofrequency thermocoagulation of a language-positive site experienced a language decline, compared with only 11% (3/28) who declined following radiofrequency thermocoagulation of language-negative sites. The likelihood of language decline was increased by 10-fold when radiofrequency thermocoagulation included a language-positive site/s compared with patients in whom no language-positive sites were coagulated. In contrast, decline was not associated with age at radiofrequency thermocoagulation, age at epilepsy diagnosis, premorbid intellectual function, number of coagulation sites or radiofrequency thermocoagulation within the dominant hemisphere. This study shows that small nodes within language networks can be essential to support function. Moreover, the premorbid integrity or 'functional adequacy' of cognitive networks might determine the capacity to compensate effectively for radiofrequency thermocoagulation of language-positive sites. These findings reveal new intricacies to network organization of cognitive functions in epilepsy and highlight the clinical advantages of language mapping for identifying patients at risk of decline following radiofrequency thermocoagulation.

A relationship between migraine without aura (MO) and patent foramen ovale (PFO) has been observed, but the neural basis underlying this relationship remains elusive. Utilizing independent component analysis via functional magnetic resonance imaging, we examined functional connectivity (FC) within and across networks in 146 patients with MO (75 patients with and 71 patients without PFO) and 70 healthy controls (35 patients each with and without PFO) to elucidate the individual effects of MO and PFO, as well as their interaction, on brain functional networks. The main effect of PFO manifested exclusively in the FC among the visual, auditory, default mode, dorsal attention and salience networks. Furthermore, the interaction effect between MO and PFO was discerned in brain clusters of the left frontoparietal network and lingual gyrus network, as well as the internetwork FC between the left frontoparietal network and the default mode network (DMN), the occipital pole and medial visual networks, and the dorsal attention and salience networks. Our findings suggest that the presence of a PFO shunt in patients with MO is accompanied by various FC changes within and across networks. These changes elucidate the intricate mechanisms linked to PFO-associated migraines and provide a basis for identifying novel noninvasive biomarkers.

INTRODUCTION:The current study examined the contributions of comprehensive neuropsychological assessment and volumetric assessment of selected mesial temporal subregions on structural magnetic resonance imaging (MRI) to identify patients with amnestic mild cognitive impairment (aMCI) and mild probable Alzheimer's disease (AD) dementia in a memory clinic cohort. METHODS:Comprehensive neuropsychological assessment and automated entorhinal, transentorhinal, and hippocampal volume measurements were conducted in 40 healthy controls, 38 patients with subjective memory symptoms, 16 patients with aMCI, 16 patients with mild probable AD dementia. Multinomial logistic regression was used to compare the neuropsychological and MRI measures. RESULTS:Combining the neuropsychological and MRI measures improved group membership prediction over the MRI measures alone but did not improve group membership prediction over the neuropsychological measures alone. CONCLUSION:Comprehensive neuropsychological assessment was an important tool to evaluate cognitive impairment. The mesial temporal volumetric MRI measures contributed no diagnostic value over and above the determinations made through neuropsychological assessment.

In children, objective, quantitative tools that determine functional neurodevelopment are scarce and rarely scalable for clinical use. Direct recordings of cortical activity using routinely acquired electroencephalography (EEG) offer physiologically reliable measures of brain function. Here, we develop a novel measure of functional brain age (FBA) using a residual neural network based interpretation of the pediatric EEG. We show that the FBA from a 10 to 15 minute segment of 18-channel EEG during light sleep (stages 1 and 2) in typically developing children and adolescents was strongly associated with chronological age (R 2 = 0.96, 95%CI: 0.94 - 0.96, n = 1062, age range: 1 month to 18 years). The mean absolute error (MAE) between FBA and age was 0.6 years ( n = 1062), with an MAE of 2.1 years following validation on an independent set of EEG recordings ( n = 723). The FBA detected group level maturational delays in a small cohort of children with abnormal neurodevelopment ( p = 0.00053, n = 40). Our work offers a practical, scalable and powerful automated tool for tracking maturation of brain function throughout childhood with an accuracy comparable to that of widely used physical growth charts.

SUMMARY:EEG source imaging (ESI) has gained traction in recent years as a useful clinical tool for the noninvasive surgical work-up of patients with drug-resistant focal epilepsy. Despite its proven benefits for the temporo-spatial modeling of spike and seizure sources, ESI remains widely underused in clinical practice. This partly relates to a lack of clarity around an optimal approach to the acquisition and processing of scalp EEG data for the purpose of ESI. Here, we describe some of the practical considerations for the clinical application of ESI. We focus on patient preparation, the impact of electrode number and distribution across the scalp, the benefit of averaging raw data for signal analysis, and the relevance of modeling different phases of the interictal discharge as it evolves from take-off to peak. We emphasize the importance of recording high signal-to-noise ratio data for reliable source analysis. We argue that the accuracy of modeling cortical sources can be improved using higher electrode counts that include an inferior temporal array, by averaging interictal waveforms rather than limiting ESI to single spike analysis, and by careful interrogation of earlier phase components of these waveforms. No amount of postacquisition signal processing or source modeling sophistication, however, can make up for suboptimally recorded scalp EEG data in a poorly prepared patient.

Surgical resection for epilepsy often fails due to incomplete Epileptogenic Zone Network (EZN) localization from scalp electroencephalography (EEG), stereo-EEG (SEEG), and Magnetic Resonance Imaging (MRI). Subjective interpretation based on interictal, or ictal recordings limits conventional EZN localization. This study employs multimodal analysis using high-density-EEG (HDEEG), Magnetoencephalography (MEG), functional-MRI (fMRI), and SEEG to overcome these limitations in a patient with drug-resistant MRI-negative focal epilepsy. A 17-year-old with drug-resistant epilepsy underwent evaluation. HDEEG, MEG, fMRI, and SEEG were used, with a novel HDEEG-cap facilitating simultaneous EEG-MEG and EEG-fMRI recordings. Electrical and magnetic source imaging were performed, and fMRI data were analysed for homogenous regions. SEEG analysis involved spike detection, spike timing analysis, ictal fast activity quantification, and Granger-based connectivity analysis. Non-invasive sessions revealed consistent interictal source imaging results identifying the EZN in the right anterior cingulate cortex. EEG-fMRI highlighted broader activation in the right cingulate cortex. SEEG analysis localized spikes and fast activity in the right anterior and posterior cingulate gyri. Multi-modal analysis suggested the EZN in the right frontal lobe, primarily involving the anterior and mid-cingulate cortices. Multi-modal non-invasive analyses can optimise SEEG implantation and surgical decision-making. Invasive analyses corroborated non-invasive findings, emphasising the importance of individual-case quantitative analysis across modalities in complex epilepsy cases.

Objective: The neuropsychological profile of patients with psychosis of epilepsy (POE) has received limited research attention. Recent neuroimaging work in POE has identified structural network pathology in the default mode network and the cognitive control network. This study examined the neuropsychological profile of POE focusing on cognitive domains subserved by these networks.Methods: Twelve consecutive patients with a diagnosis of POE were prospectively recruited from the Comprehensive Epilepsy Programmes at The Royal Melbourne, Austin and St Vincent's Hospitals, Melbourne, Australia between January 2015 and February 2017. They were compared to 12 matched patients with epilepsy but no psychosis and 42 healthy controls on standardised neuropsychological tests of memory and executive functioning in a case-control design.Results: Mean scores across all cognitive tasks showed a graded pattern of impairment, with the POE group showing the poorest performance, followed by the epilepsy without psychosis and the healthy control groups. This was associated with significant group-level differences on measures of working memory (p = < 0.01); immediate (p = < 0.01) and delayed verbal recall (p = < 0.01); visual memory (p < 0.001); and verbal fluency (p = 0.02). In particular, patients with POE performed significantly worse than the healthy control group on measures of both cognitive control (p = .005) and memory (p < .001), whereas the epilepsy without psychosis group showed only memory difficulties (delayed verbal recall) compared to healthy controls (p = .001).Conclusion: People with POE show reduced performance in neuropsychological functions supported by the default mode and cognitive control networks, when compared to both healthy participants and people with epilepsy without psychosis.

Objective. Magnetoencephalography (MEG) is a powerful non-invasive diagnostic modality for presurgical epilepsy evaluation. However, the clinical utility of MEG mapping for localising epileptic foci is limited by its low efficiency, high labour requirements, and considerable interoperator variability. To address these obstacles, we proposed a novel artificial intelligence–based automated magnetic source imaging (AMSI) pipeline for automated detection and localisation of epileptic sources from MEG data. Approach. To expedite the analysis of clinical MEG data from patients with epilepsy and reduce human bias, we developed an autolabelling method, a deep-learning model based on convolutional neural networks and a hierarchical clustering method based on a perceptual hash algorithm, to enable the coregistration of MEG and magnetic resonance imaging, the detection and clustering of epileptic activity, and the localisation of epileptic sources in a highly automated manner. We tested the capability of the AMSI pipeline by assessing MEG data from 48 epilepsy patients. Main results. The AMSI pipeline was able to rapidly detect interictal epileptiform discharges with 93.31% ± 3.87% precision based on a 35-patient dataset (with sevenfold patientwise cross-validation) and robustly rendered accurate localisation of epileptic activity with a lobar concordance of 87.18% against interictal and ictal stereo-electroencephalography findings in a 13-patient dataset. We also showed that the AMSI pipeline accomplishes the necessary processes and delivers objective results within a much shorter time frame (∼12 min) than traditional manual processes (∼4 h). Significance. The AMSI pipeline promises to facilitate increased utilisation of MEG data in the clinical analysis of patients with epilepsy.

The current study compared the reliability of manual collateral sulcus depth and entorhinal and transentorhinal cortical volume measurements between native oriented MRI scans versus MRI scans realigned to the hippocampal long axis. Data included 10 participants with two serial 3.0T MRI scans from the Alzheimer's Disease Neuroimaging Initiative. Both collateral sulcus depth and entorhinal and transentorhinal cortical volume measurement reliability improved from the native to the hippocampal oriented scans. Standardizing scan orientation is important to optimize reliability of MRI-derived manual measurements of the entorhinal and transentorhinal cortices. In quantitative MRI studies, aligning scans to a common, normalized orientation is recommended.

Alzheimer’s disease (AD) can be challenging to diagnose early as cognitive changes are mild. Neuropsychological assessment, in contrast to cognitive screening or isolated cognitive tests, are more sensitive to the mild cognitive changes in early AD. Magnetic resonance imaging (MRI) can detect early structural brain changes in AD, particularly entorhinal cortex, transentorhinal cortex, and hippocampal atrophy. Combining neuropsychological assessment with MRI examination may yield improved ability to detect early AD. The current study examined combining neuropsychological and MRI measures to detect patients with mild cognitive impairment (MCI) in a memory clinic sample. Neuropsychological assessment and entorhinal cortex, transentorhinal cortex, and hippocampal volume measurement were conducted in memory clinic patients, including 38 with memory symptoms, 16 with MCI, and 16 with dementia due to AD, and in 40 healthy controls. Backward stepwise multinomial logistic regression was used to compare the neuropsychological measures and MRI measures. Among the neuropsychological measures, memory and processing speed were the most important measures that contributed to group membership prediction. All MRI measures contributed to group membership prediction. The neuropsychological measures were more accurate than the MRI measures at detecting patients with MCI. Combining the MRI measures with the neuropsychological measures did not improve group membership prediction. Neuropsychological assessment is an important tool to support MCI diagnosis. MRI measures contribute little diagnostic value over and above neuropsychological assessment. Further advancement in deriving MRI-based measurements is required to achieve clinical utility.

The current study aimed to validate entorhinal and transentorhinal cortical volumes measured by the automated segmentation tool Automatic Segmentation of Hippocampal Subfields (ASHS-T1). The study sample comprised 34 healthy controls (HCs), 37 individuals with amnestic mild cognitive impairment (aMCI), and 29 individuals with Alzheimer's disease (AD) dementia from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Entorhinal and transentorhinal cortical volumes were assessed using ASHS-T1, manual segmentation, as well as a widely used automated segmentation tool, FreeSurfer v6.0.1. Mean differences, intraclass correlation coefficients, and Bland-Altman plots were computed. ASHS-T1 tended to underestimate entorhinal and transentorhinal cortical volumes relative to manual segmentation and FreeSurfer. There was variable consistency and low agreement between ASHS-T1 and manual segmentation volumes. There was low-to-moderate consistency and low agreement between ASHS-T1 and FreeSurfer volumes. There was a trend toward higher consistency and agreement for the entorhinal cortex in the aMCI and AD groups compared to the HC group. Despite the differences in volume measurements, ASHS-T1 was sensitive to entorhinal and transentorhinal cortical atrophy in both early and late disease stages. Based on the current study, ASHS-T1 appears to be a promising tool for automated entorhinal and transentorhinal cortical volume measurement in individuals with likely underlying AD.

Closed-loop brain-computer interface (BCI) systems that provide real-time feedback to their users are essential for the synthesis of attempted or imagined speech from intracranial recordings. Here, we describe the implementation of our BCI speech synthesis system, which can be trained with a limited amount of overt speech to produce a continuous stream of audio outputs during subsequent speech imagery tasks. We evaluate (1) the effect of parameter choices on the execution time of individual operations in the BCI loop and (2) the accuracy of predicted outputs. To confirm the feasibility of our approach, we conduct simulations in a pseudo-prospective fashion using recorded datasets from five patients undergoing intracranial epilepsy monitoring. We propose that our system can be used to synthesize different types of speech under specific clinical constraints.

Background Automated magnetic resonance imaging (MRI) volumetry is a promising tool to evaluate regional brain volumes in dementia and especially Alzheimer's disease (AD). Purpose To compare automated methods and the gold standard manual segmentation in measuring regional brain volumes on MRI across healthy controls, patients with mild cognitive impairment, and patients with dementia due to AD. Study Type Systematic review and meta-analysis. Data Sources MEDLINE, Embase, and PsycINFO were searched through October 2021. Field Strength 1.0 T, 1.5 T, or 3.0 T. Assessment Two review authors independently identified studies for inclusion and extracted data. Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Statistical Tests Standardized mean differences (SMD; Hedges' g) were pooled using random-effects meta-analysis with robust variance estimation. Subgroup analyses were undertaken to explore potential sources of heterogeneity. Sensitivity analyses were conducted to examine the impact of the within-study correlation between effect estimates on the meta-analysis results. Results Seventeen studies provided sufficient data to evaluate the hippocampus, lateral ventricles, and parahippocampal gyrus. The pooled SMD for the hippocampus, lateral ventricles, and parahippocampal gyrus were 0.22 (95% CI -0.50 to 0.93), 0.12 (95% CI -0.13 to 0.37), and -0.48 (95% CI -1.37 to 0.41), respectively. For the hippocampal data, subgroup analyses suggested that the pooled SMD was invariant across clinical diagnosis and field strength. Subgroup analyses could not be conducted on the lateral ventricles data and the parahippocampal gyrus data due to insufficient data. The results were robust to the selected within-study correlation value. Data Conclusion While automated methods are generally comparable to manual segmentation for measuring hippocampal, lateral ventricle, and parahippocampal gyrus volumes, wide 95% CIs and large heterogeneity suggest that there is substantial uncontrolled variance. Thus, automated methods may be used to measure these regions in patients with AD but should be used with caution. Evidence Level 3 Technical Efficacy Stage 3

Objective: To explore the cortical morphological associations of the psychoses of epilepsy. Methods: Psychosis of epilepsy (POE) has two main subtypes - postictal psychosis and interictal psychosis. We used automated surface-based analysis of magnetic resonance images to compare cortical thickness, area, and volume across the whole brain between: (i) all patients with POE (n = 23) relative to epilepsy-without psychosis controls (EC; n = 23), (ii) patients with interictal psychosis (n = 10) or postictal psychosis (n = 13) relative to EC, and (iii) patients with postictal psychosis (n = 13) relative to patients with interictal psychosis (n = 10). Results: POE is characterised by cortical thickening relative to EC, occurring primarily in nodes of the cognitive control network; (rostral anterior cingulate, caudal anterior cingulate, middle frontal gyrus), and the default mode network (posterior cingulate, medial paracentral gyrus, and precuneus). Patients with interictal psychosis displayed cortical thickening in the left hemisphere in occipital and temporal regions relative to EC (lateral occipital cortex, lingual, fusiform, and inferior temporal gyri), which was evident to a lesser extent in postictal psychosis patients. There were no significant differences in cortical thickness, area, or volume between the postictal psychosis and EC groups, or between the postictal psychosis and interictal psychosis groups. However, prior to correction for multiple comparisons, both the interictal psychosis and postictal psychosis groups displayed cortical thickening relative to EC in highly similar regions to those identified in the POE group overall. Significance: The results show cortical thickening in POE overall, primarily in nodes of the cognitive control and default mode networks, compared to patients with epilepsy without psychosis. Additional thickening in temporal and occipital neocortex implicated in the dorsal and ventral visual pathways may differentiate interictal psychosis from postictal psychosis. A novel mechanism for cortical thickening in POE is proposed whereby normal synaptic pruning processes are interrupted by seizure onset.

There is an urgent need for more informative quantitative techniques that non-invasively and objectively assess strategies for epilepsy surgery. Invasive intracranial electroencephalography (iEEG) remains the clinical gold standard to investigate the nature of the epileptogenic zone (EZ) before surgical resection. However, there are major limitations of iEEG, such as the limited spatial sampling and the degree of subjectivity inherent in the analysis and clinical interpretation of iEEG data. Recent advances in network analysis and dynamical network modeling provide a novel aspect toward a more objective assessment of the EZ. The advantage of such approaches is that they are data-driven and require less or no human input. Multiple studies have demonstrated success using these approaches when applied to iEEG data in characterizing the EZ and predicting surgical outcomes. However, the limitations of iEEG recordings equally apply to these studies—limited spatial sampling and the implicit assumption that iEEG electrodes, whether strip, grid, depth or stereo EEG (sEEG) arrays, are placed in the correct location. Therefore, it is of interest to clinicians and scientists to see whether the same analysis and modeling techniques can be applied to whole-brain, non-invasive neuroimaging data (from MRI-based techniques) and neurophysiological data (from MEG and scalp EEG recordings), thus removing the limitation of spatial sampling, while safely and objectively characterizing the EZ. This review aims to summarize current state of the art non-invasive methods that inform epilepsy surgery using network analysis and dynamical network models. We also present perspectives on future directions and clinical applications of these promising approaches.

We developed a voice-based, self-paced cursor control task to collect corresponding intracranial neural data during isolated utterances of phonemes, namely vowel, nasal and fricative sounds. Two patients implanted with intracranial depth electrodes for clinical epilepsy monitoring performed closed-loop voice-based cursor control from real-time processing of microphone input. In post-hoc data analyses, we searched for neural features that correlated with the occurrence of non-specific speech sounds or specific phonemes. In line with previous studies, we observed onset and sustained responses to speech sounds at multiple recording sites within the superior temporal gyrus. Based on differential patterns of activation in narrow frequency bands up to 200 Hz, we tracked voice activity with 91% accuracy (chance level: 50%) and classified individual utterances into one of five phonemes with 68% accuracy (chance level: 20%). We propose that our framework could be extended to additional phonemes to better characterize neurophysiological mechanisms underlying the production and perception of speech sounds in the absence of language context. In general, our findings provide supplementary evidence and information toward the development of speech brain-computer interfaces using intracranial electrodes.

Modelling the interactions that arise from neural dynamics in seizure genesis is challenging but important in the effort to improve the success of epilepsy surgery. Dynamical network models developed from physiological evidence offer insights into rapidly evolving brain networks in the epileptic seizure. A limitation of previous studies in this field is the dependence on invasive cortical recordings with constrained spatial sampling of brain regions that might be involved in seizure dynamics. Here, we propose virtual intracranial electroencephalography (ViEEG), which combines non-invasive ictal magnetoencephalographic imaging (MEG), dynamical network models and a virtual resection technique. In this proof-of-concept study, we show that ViEEG signals reconstructed from MEG alone preserve critical temporospatial characteristics for dynamical approaches to identify brain areas involved in seizure generation. We show the non-invasive ViEEG approach may have some advantage over intracranial electroencephalography (iEEG). Future work may be designed to test the potential of the virtual iEEG approach for use in surgical management of epilepsy.

Objectives: To determine the long-term outcomes in patients undergoing intracranial EEG (iEEG) evaluation for epilepsy surgery in terms of seizure freedom, mood, and quality of life at St. Vincent's Hospital, Melbourne. Methods: Patients who underwent iEEG between 1999 and 2016 were identified. Patients were retrospectively assessed between 2014 and 2017 by specialist clinic record review and telephone survey with standardized validated questionnaires for: 1) seizure freedom using the Engel classification; 2) Mood using the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E); 3) Quality-of-life outcomes using the QOLIE-10 questionnaire. Summary statistics and univariate analysis were performed to investigate variables for significance. Results: Seventy one patients underwent iEEG surgery: 49 Subdural, 14 Depths, 8 Combination with 62/68 (91.9%) of those still alive, available at last follow-up by telephone survey or medical record review (median of 8.2 years). The estimated epileptogenic zone was 62% temporal and 38% extra-temporal. At last follow-up, 69.4% (43/62) were Engel Class I and 30.6% (19/62) were Engel Class II-IV. Further, a depressive episode (NDDI-E > 15) was observed in 34% (16/47), while a 'better quality of life' (QOLIE10 score < 25) was noted in 74% (31/42). Quality of life (p < 0.001) but not mood (p = 0.24) was associated with seizure freedom. Significance: Long-term seizure freedom can be observed in patients undergoing complex epilepsy surgery with iEEG evaluation and is associated with good quality of life. (c) 2021 Elsevier Inc. All rights reserved.

Speech imagery is a mental strategy that paralyzed patients can use to control a brain-computer interface (BCI) at their own pace. Most studies that have attempted to decode speech have used scalp electroencephalography or electrocorticography. Only few studies have used stereotactic electroencephalography (SEEG), which enables the exploration of deeply located structures in the brain, in this context. In this paper, we aim to identify discriminative features for decoding speech perception and overt and imagined speech production from SEEG recordings in three patients with epilepsy. We report results for the detection of speech events and for the classification of the corresponding utterances. We propose that SEEG-based BCI systems with multiple degrees of freedom may be reliably controlled by selected phonetic features decoded from the superior temporal gyrus.