XDL-ESI: Electrophysiological Sources Imaging via explainable deep learning framework with validation on simultaneous EEG and iEEG.

Electroencephalography (EEG) or Magnetoencephalography (MEG) source imaging aims to estimate the underlying activated brain sources to explain the observed EEG/MEG recordings. Solving the inverse problem of EEG/MEG Source Imaging (ESI) is challenging due to its ill-posed nature. To achieve a unique solution, it is essential to apply sophisticated regularization constraints to restrict the solution space. Traditionally, the design of regularization terms is based on assumptions about the spatiotemporal structure of the underlying source dynamics. In this paper, we propose a novel paradigm for ESI via an Explainable Deep Learning framework, termed as XDL-ESI, which connects the iterative optimization algorithm with deep learning architecture by unfolding the iterative updates with neural network modules. The proposed framework has the advantages of (1) establishing a data-driven approach to model the source solution structure instead of using hand-crafted regularization terms; (2) improving the robustness of source solutions by introducing a topological loss that leverages the geometric spatial information applying varying penalties on distinct localization errors; (3) improving the reconstruction efficiency and interpretability as it inherits the advantages from both the iterative optimization algorithms (interpretability) and deep learning approaches (function approximation). The proposed XDL-ESI framework provides an efficient, accurate, and interpretable paradigm to solve the ESI inverse problem with satisfactory performance in both simulated data and real clinical data. Specially, this approach is further validated using simultaneous EEG and intracranial EEG (iEEG).

Accuracy of omni-planar and surface casting of epileptiform activity for intracranial seizure localization.

OBJECTIVE: Intracranial electroencephalography (ICEEG) recordings are performed for seizure localization in medically refractory epilepsy. Signal quantifications such as frequency power can be projected as heatmaps on personalized three-dimensional (3D) reconstructed cortical surfaces to distill these complex recordings into intuitive cinematic visualizations. However, simultaneously reconciling deep recording locations and reliably tracking evolving ictal patterns remain significant challenges. METHODS: We fused oblique magnetic resonance imaging (MRI) slices along depth probe trajectories with cortical surface reconstructions and projected dynamic heatmaps using a simple mathematical metric of epileptiform activity (line-length). This omni-planar and surface casting of epileptiform activity approach (OPSCEA) thus illustrated seizure onset and spread among both deep and superficial locations simultaneously with minimal need for signal processing supervision. We utilized the approach on 41 patients at our center implanted with grid, strip, and/or depth electrodes for localizing medically refractory seizures. Peri-ictal data were converted into OPSCEA videos with multiple 3D brain views illustrating all electrode locations. Five people of varying expertise in epilepsy (medical student through epilepsy attending level) attempted to localize the seizure-onset zones. RESULTS: We retrospectively compared this approach with the original ICEEG study reports for validation. Accuracy ranged from 73.2% to 97.6% for complete or overlapping onset lobe(s), respectively, and ~56.1% to 95.1% for the specific focus (or foci). Higher answer certainty for a given case predicted better accuracy, and scorers had similar accuracy across different training levels. SIGNIFICANCE: In an era of increasing stereo-EEG use, cinematic visualizations fusing omni-planar and surface functional projections appear to provide a useful adjunct for interpreting complex intracranial recordings and subsequent surgery planning.

Automatic wavelet-based assessment of behavioral sleep using multichannel electrocorticography in rats.

PURPOSE: During the last decade, the reported prevalence of sleep-disordered breathing in adults has been rapidly increasing. Therefore, automatic methods of sleep assessment are of particular interest. In a framework of translational neuroscience, this study introduces a reliable automatic detection system of behavioral sleep in laboratory rats based on the signal recorded at the cortical surface without requiring electromyography. METHODS: Experimental data were obtained in 16 adult male WAG/Rij rats at the age of 9 months. Electrocorticographic signals (ECoG) were recorded in freely moving rats during the entire day (22.5 ± 2.2 h). Automatic wavelet-based assessment of behavioral sleep (BS) was proposed. The performance of this wavelet-based method was validated in a group of rats with genetic predisposition to absence epilepsy (n=16) based on visual analysis of their behavior in simultaneously recorded video. RESULTS: The accuracy of automatic sleep detection was 98% over a 24-h period. An automatic BS assessment method can be adjusted for detecting short arousals during sleep (microarousals) with various duration. CONCLUSIONS: These findings suggest that automatic wavelet-based assessment of behavioral sleep can be used for assessment of sleep quality. Current analysis indicates a temporal relationship between microarousals, sleep, and epileptic discharges in genetically prone subjects.

Value and limitations of intracranial recordings for validating electric field modeling for transcranial brain stimulation.

Comparing electric field simulations from individualized head models against in-vivo intra-cranial recordings is considered the gold standard for direct validation of computational field modeling for transcranial brain stimulation and brain mapping techniques such as electro- and magnetoencephalography. The measurements also help to improve simulation accuracy by pinning down the factors having the largest influence on the simulations. Here we compare field simulations from four different automated pipelines against intracranial voltage recordings in an existing dataset of 14 epilepsy patients. We show that modeling differences in the pipelines lead to notable differences in the simulated electric field distributions that are often large enough to change the conclusions regarding the dose distribution and strength in the brain. Specifically, differences in the automatic segmentations of the head anatomy from structural magnetic resonance images are a major factor contributing to the observed field differences. However, the differences in the simulated fields are not reflected in the comparison between the simulations and intra-cranial measurements. This apparent mismatch is partly explained by the noisiness of the intra-cranial measurements, which renders comparisons between the methods inconclusive. We further demonstrate that a standard regression analysis, which ignores uncertainties in the simulations, leads to a strong bias in the estimated linear relationship between simulated and measured fields. Ignoring this bias leads to the incorrect conclusion that the models systematically misestimate the field strength in the brain. We propose a new Bayesian regression analysis of the data that yields unbiased parameter estimates, along with their uncertainties, and gives further insights to the fit between simulations and measurements. Specifically, the unbiased results give only weak support for systematic misestimations of the fields by the models.

Low entropy map of brain oscillatory activity identifies spatially localized events: A new method for automated epilepsy focus prediction.

The spatial mapping of localized events in brain activity critically depends on the correct identification of the pattern signatures associated with those events. For instance, in the context of epilepsy research, a number of different electrophysiological patterns have been associated with epileptogenic activity. Motivated by the need to define automated seizure focus detectors, we propose a novel data-driven algorithm for the spatial identification of localized events that is based on the following rationale: the distribution of emerging oscillations during confined events across all recording sites is highly non-uniform and can be mapped using a spatial entropy function. By applying this principle to EEG recording obtained from 67 distinct seizure epochs, our method successfully identified the seizure focus on a group of ten drug-resistant temporal lobe epilepsy patients (average sensitivity: 0.94, average specificity: 0.90) together with its characteristic electrophysiological pattern signature. Cross-validation of the method outputs with postresective information revealed the consistency of our findings in long follow-up seizure-free patients. Overall, our methodology provides a reliable computational procedure that might be used as in both experimental and clinical domains to identify the neural populations undergoing an emerging functional or pathological transition.

Cortical hemodynamic responses induced by low-intensity transcranial ultrasound stimulation of mouse cortex.

Ultrasound-mediated neuromodulation is emerging as a key technology for targeted noninvasive brain stimulation, but key insights into its effects and dose-response characteristics are still missing. The purpose of this study is to systematically evaluate the effect of low-intensity transcranial ultrasound stimulation (TUS) on complementary aspects of cerebral hemodynamic. We simultaneously record the EMG signal, local field potential (LFP) and cortical blood flow (CBF) using electrophysiological recording and laser speckle contrast imaging under ultrasound stimulation to simultaneously monitor motor responses, neural activities and hemodynamic changes during the application of low-intensity TUS in mouse motor cortex, using excitation pulses which caused whisker and tail movement. Our experimental results demonstrate interdependent TUS-induced motor, neural activity and hemodynamic responses that peak approximately 0.55s, 1.05s and 2.5s after TUS onset, respectively, and show a linear coupling relationship between their respective varying response amplitudes to repeated stimuli. We also found monotonic dose-response parametric relations of the CBF peak value increase as a function of stimulation intensity and duration, while stimulus duty-cycle had only a weak effect on peak responses. These findings demonstrate that TUS induces a change in cortical hemodynamics and LSCI provide a high temporal resolution view of these changes.

Real-time presurgical resting-state fMRI in patients with brain tumors: Quality control and comparison with task-fMRI and intraoperative mapping.

Resting-state functional magnetic resonance imaging (rsfMRI) is a promising task-free functional imaging approach, which may complement or replace task-based fMRI (tfMRI) in patients who have difficulties performing required tasks. However, rsfMRI is highly sensitive to head movement and physiological noise, and validation relative to tfMRI and intraoperative electrocortical mapping is still necessary. In this study, we investigate (a) the feasibility of real-time rsfMRI for presurgical mapping of eloquent networks with monitoring of data quality in patients with brain tumors and (b) rsfMRI localization of eloquent cortex compared with tfMRI and intraoperative electrocortical stimulation (ECS) in retrospective analysis. Five brain tumor patients were studied with rsfMRI and tfMRI on a clinical 3T scanner using MultiBand(8)-echo planar imaging (EPI) with repetition time: 400 ms. Moving-averaged sliding-window correlation analysis with regression of motion parameters and signals from white matter and cerebrospinal fluid was used to map sensorimotor and language resting-state networks. Data quality monitoring enabled rapid optimization of scan protocols, early identification of task noncompliance, and head movement-related false-positive connectivity to determine scan continuation or repetition. Sensorimotor and language resting-state networks were identifiable within 1 min of scan time. The Euclidean distance between ECS and rsfMRI connectivity and task-activation in motor cortex, Broca's, and Wernicke's areas was 5-10 mm, with the exception of discordant rsfMRI and ECS localization of Wernicke's area in one patient due to possible cortical reorganization and/or altered neurovascular coupling. This study demonstrates the potential of real-time high-speed rsfMRI for presurgical mapping of eloquent cortex with real-time data quality control, and clinically acceptable concordance of rsfMRI with tfMRI and ECS localization.

Preictal variability of high-frequency oscillation rates in refractory epilepsy.

OBJECTIVE: High-frequency oscillations (HFOs) have shown promising utility in the spatial localization of the seizure onset zone for patients with focal refractory epilepsy. Comparatively few studies have addressed potential temporal variations in HFOs, or their role in the preictal period. Here, we introduce a novel evaluation of the instantaneous HFO rate through interictal and peri-ictal epochs to assess their usefulness in identifying imminent seizure onset. METHODS: Utilizing an automated HFO detector, we analyzed intracranial electroencephalographic data from 30 patients with refractory epilepsy undergoing long-term presurgical evaluation. We evaluated HFO rates both as a 30-minute average and as a continuous function of time and used nonparametric statistical methods to compare individual and population-level differences in rate during peri-ictal and interictal periods. RESULTS: Mean HFO rate was significantly higher for all epochs in seizure onset zone channels versus other channels. Across the 30 patients of our cohort, we found no statistically significant differences in mean HFO rate during preictal and interictal epochs. For continuous HFO rates in seizure onset zone channels, however, we found significant population-wide increases in preictal trends relative to interictal periods. Using a data-driven analysis, we identified a subset of 11 patients in whom either preictal HFO rates or their continuous trends were significantly increased relative to those of interictal baseline and the rest of the population. SIGNIFICANCE: These results corroborate existing findings that HFO rates within epileptic tissue are higher during interictal periods. We show this finding is also present in preictal, ictal, and postictal data, and identify a novel biomarker of preictal state: an upward trend in HFO rate leading into seizures in some patients. Overall, our findings provide preliminary evidence that HFOs can function as a temporal biomarker of seizure onset.

Comparison of narcotic pain control between stereotactic electrocorticography and subdural grid implantation.

INTRODUCTION: The choice of subdural grid (SDG) or stereoelectroencephalography (sEEG) for patients with epilepsy can be complex and in some cases overlap. Comparing postoperative pain and narcotics consumption with SDG or sEEG can help develop an intracranial monitoring strategy. MATERIALS AND METHODS: A retrospective study was performed for adult patients undergoing SDG or sEEG monitoring. Numeric Rating Scale (NRS) was used for pain assessment. Types and dosage of the opioids were calculated by converting into milligram morphine equivalents (MME). Narcotic consumption was analyzed at the following three time periods: I. the first 24 h of implantation; II. from the second postimplantation day to the day of explantation; and III. the days following electrode removal to discharge. RESULTS: Forty-two patients who underwent SDG and 31 patients who underwent sEEG implantation were analyzed. After implantation, average NRS was 3.7 for SDG and 2.2 for sEEG (P < .001). After explantation, the NRS was 3.5 for SDG and 1.4 in sEEG (P < .001). Sixty percent of SDG patients and 13% of sEEG patients used more than one opioid in period III (P < .001). The SDG group had a significantly higher MME throughout the three periods compared with the sEEG group: period I: 448 (SDG) vs. 205 (sEEG) mg, P = .002; period II: 377 (SDG) vs. 102 (sEEG) mg, P < .001; and period III: 328 (SDG) vs. 75 (sEEG) mg; P = .002. Patients with the larger SDG implantation had the higher NRS (P = .03) and the higher MME at period I (P = .019). There was no correlation between the number of depth electrodes and pain control in patients with sEEG. CONCLUSIONS: Patients undergoing sEEG had significantly less pain and required fewer opiates compared with patients with SDG. These differences in perioperative pain may be a consideration when choosing between these two invasive monitoring options.

Decoding spectro-temporal representation for motor imagery recognition using ECoG-based brain-computer interfaces.

One of the challenges in brain-computer interface systems is obtaining motor imagery recognition from brain activities. Brain-signal decoding robustness and system performance improvement during the motor imagery process are two of the essential issues in brain-computer interface research. In conventional approaches, ineffective decoding of features and high complexity of algorithms often lead to unsatisfactory performance. A novel method for the recognition of motor imagery tasks is developed based on employing a modified S-transforms for spectro-temporal representation to characterize the behavior of electrocorticogram activities. A classifier is trained by using a support vector machine, and an optimized wrapper approach is applied to guide selection to implement the representation selection obtained. A channel selection algorithm optimizes the wrapper approach by adding a cross-validation step, which effectively improves the classification performance. The modified S-transform can accurately capture event-related desynchronization/event-related synchronization phenomena and can effectively locate sensorimotor rhythm information. The optimized wrapper approach used in this scheme can effectively reduce the feature dimension and improve algorithm efficiency. The method is evaluated on a public electrocorticogram dataset with a recognition accuracy of 98% and an information transfer rate of 0.8586 bit/trial. To verify the effect of the channel selection, both electrocorticogram and electroencephalogram data are experimentally analyzed. Furthermore, the computational efficiency of this scheme demonstrates its potential for online brain-computer interface systems in future cognitive tasks.

Current value of intraoperative electrocorticography (iopECoG).

OBJECTIVE: Intraoperative electrocorticography (iopECoG) can contribute to delineate the resection borders of the anticipated epileptogenic zone in epilepsy surgery. However, it has several caveats that should be considered to avoid incorrect interpretation during intraoperative monitoring. METHODS: The literature on iopECoG application was reviewed, and pros and cons as well as obstacles to this technique were analyzed. RESULTS: The literature of the first half of the nineties was very enthusiastic in using iopECoG for tailoring the resection in temporal as well as extratemporal epilepsy surgery. Mostly, this resulted in a good correlation of postresection ECoG and excellent seizure outcome. In the second half of the nineties, many authors demonstrated lack of correlation between iopECoG and postoperative seizure outcome, especially in surgery for temporal lobe epilepsy with hippocampal sclerosis. In the noughties, investigators found that ECoG was significantly useful in neocortical lesional temporal lobe epilepsy as well as in extratemporal lesional epilepsies. Extratemporal epilepsy without lesions proved to be more a domain of chronic extraoperative ECoG, especially using depth electrode recordings. In recent years, iopECoG detecting high-frequency oscillations (ripples, 80-250 Hz, fast ripples, 250-500 Hz) for tailored resection was found to allow intraoperative prediction of postoperative seizure outcome. CONCLUSION: After a period of scepticism, iopECoG seems back in the focus of interest for intraoperative guidance of resecting epileptogenic tissue to raise postoperative favorable seizure outcome. In temporal and extratemporal lesional epilepsies, especially in cases of focal cortical dysplasia, tuberous sclerosis, or cavernous malformations, an excellent correlation between iopECoG-guided resection and postoperative seizure relief was found.

Utility of stereo-electroencephalography recording guided by magnetoencephalography in the surgical treatment of epilepsy patients with negative magnetic resonance imaging results.

Objective: It is challenging for neurosurgeons to perform surgeries on patients without detectable structural lesions. Therefore, this retrospective study aimed to explore the outcome of stereo-electroencephalography (SEEG) in suspicious areas guided by magnetoencephalography (MEG)-magnetic resonance imaging (MRI) reconstruction in MRI-negative epilepsy patients. Methods: This study included 47 patients with negative-MRI epilepsy. Seizure outcome at 24 months was assessed using a modified Engel's classification. Accordingly, class I and II were considered favorable outcomes, whereas classes III and IV were unfavorable. Furthermore, patients were classified into a consistent group if the results of MEG and SEEG indicated the same area of the brain. The relationship between surgical outcome and the concordance of MEG and SEEG was analyzed. Results: A complete seizure-free condition was achieved in 22 (47%) patients. Sex, handedness, age and duration of illness were not significantly associated with seizure-free outcome (p = .187 [Pearson chi-squared test]). The number of patients with favorable outcome (Engle I and II) was as high as 68% at the time of follow-up. Furthermore, more seizure-free patients were found in the SEEG and MEG consistent group. Conclusions: SEEG is a valuable tool in the pre-evaluation for resective epilepsy surgery, particularly in negative-MRI epilepsy patients; MEG greatly facilitates localization for SEEG electrode implantation. However, none of these tools are absolutely sensitive and reliable; therefore, collecting as much information as possible is necessary to achieve satisfactory results in epilepsy surgery.

Optimal referencing for stereo-electroencephalographic (SEEG) recordings.

Stereo-electroencephalography (SEEG) is an intracranial recording technique in which depth electrodes are inserted in the brain as part of presurgical assessments for invasive brain surgery. SEEG recordings can tap into neural signals across the entire brain and thereby sample both cortical and subcortical sites. However, even though signal referencing is important for proper assessment of SEEG signals, no previous study has comprehensively evaluated the optimal referencing method for SEEG. In our study, we recorded SEEG data from 15 human subjects during a motor task, referencing them against the average of two white matter contacts (monopolar reference). We then subjected these signals to 5 different re-referencing approaches: common average reference (CAR), gray-white matter reference (GWR), electrode shaft reference (ESR), bipolar reference, and Laplacian reference. The results from three different signal quality metrics suggest the use of the Laplacian re-reference for study of local population-level activity and low-frequency oscillatory activity.

On the importance of precise electrode placement for targeted transcranial electric stimulation.

Transcranial electric stimulation (TES) is an increasingly popular method for non-invasive modulation of brain activity and a potential treatment for neuropsychiatric disorders. However, there are concerns about the reliability of its application because of variability in TES-induced intracranial electric fields across individuals. While realistic computational models offer can help to alleviate these concerns, their direct empirical validation is sparse, and their practical implications are not always clear. In this study, we combine direct intracranial measurements of electric fields generated by TES in surgical epilepsy patients with computational modeling. First, we directly validate the computational models and identify key parameters needed for accurate model predictions. Second, we derive practical guidelines for a reliable application of TES in terms of the precision of electrode placement needed to achieve a desired electric field distribution. Based on our results, we recommend electrode placement accuracy to be < 1 cm for a reliable application of TES across sessions.

MULAN: Evaluation and ensemble statistical inference for functional connectivity.

Many analysis methods exist to extract graphs of functional connectivity from neuronal networks. Confidence in the results is limited because, (i) different methods give different results, (ii) parameter setting directly influences the final result, and (iii) systematic evaluation of the results is not always performed. Here, we introduce MULAN (MULtiple method ANalysis), which assumes an ensemble based approach combining multiple analysis methods and fuzzy logic to extract graphs with the most probable structure. In order to reduce the dependency on parameter settings, we determine the best set of parameters using a genetic algorithm on simulated datasets, whose temporal structure is similar to the experimental one. After a validation step, the selected set of parameters is used to analyze experimental data. The final step cross-validates experimental subsets of data and provides a direct estimate of the most likely graph and our confidence in the proposed connectivity. A systematic evaluation validates our strategy against empirical stereotactic electroencephalography (SEEG) and functional magnetic resonance imaging (fMRI) data.

Sub-millimeter ECoG pitch in human enables higher fidelity cognitive neural state estimation.

Electrocorticography (ECoG), electrophysiological recording from the pial surface of the brain, is a critical measurement technique for clinical neurophysiology, basic neurophysiology studies, and demonstrates great promise for the development of neural prosthetic devices for assistive applications and the treatment of neurological disorders. Recent advances in device engineering are poised to enable orders of magnitude increase in the resolution of ECoG without comprised measurement quality. This enhancement in cortical sensing enables the observation of neural dynamics from the cortical surface at the micrometer scale. While these technical capabilities may be enabling, the extent to which finer spatial scale recording enhances functionally relevant neural state inference is unclear. We examine this question by employing a high-density and low impedance 400 µm pitch microECoG (µECoG) grid to record neural activity from the human cortical surface during cognitive tasks. By applying machine learning techniques to classify task conditions from the envelope of high-frequency band (70-170Hz) neural activity collected from two study participants, we demonstrate that higher density grids can lead to more accurate binary task condition classification. When controlling for grid area and selecting task informative sub-regions of the complete grid, we observed a consistent increase in mean classification accuracy with higher grid density; in particular, 400 µm pitch grids outperforming spatially sub-sampled lower density grids up to 23%. We also introduce a modeling framework to provide intuition for how spatial properties of measurements affect the performance gap between high and low density grids. To our knowledge, this work is the first quantitative demonstration of human sub-millimeter pitch cortical surface recording yielding higher-fidelity state estimation relative to devices at the millimeter-scale, motivating the development and testing of µECoG for basic and clinical neurophysiology as well as towards the realization of high-performance neural prostheses.

Crowdsourcing seizure detection: algorithm development and validation on human implanted device recordings.

There exist significant clinical and basic research needs for accurate, automated seizure detection algorithms. These algorithms have translational potential in responsive neurostimulation devices and in automatic parsing of continuous intracranial electroencephalography data. An important barrier to developing accurate, validated algorithms for seizure detection is limited access to high-quality, expertly annotated seizure data from prolonged recordings. To overcome this, we hosted a kaggle.com competition to crowdsource the development of seizure detection algorithms using intracranial electroencephalography from canines and humans with epilepsy. The top three performing algorithms from the contest were then validated on out-of-sample patient data including standard clinical data and continuous ambulatory human data obtained over several years using the implantable NeuroVista seizure advisory system. Two hundred teams of data scientists from all over the world participated in the kaggle.com competition. The top performing teams submitted highly accurate algorithms with consistent performance in the out-of-sample validation study. The performance of these seizure detection algorithms, achieved using freely available code and data, sets a new reproducible benchmark for personalized seizure detection. We have also shared a 'plug and play' pipeline to allow other researchers to easily use these algorithms on their own datasets. The success of this competition demonstrates how sharing code and high quality data results in the creation of powerful translational tools with significant potential to impact patient care.

Assessing expert reliability in determining intracranial EEG channel quality and introducing the automated bad channel detection algorithm.

Objective.To evaluate the inter- and intra-rater reliability for the identification of bad channels among neurologists, EEG Technologists, and naïve research personnel, and to compare their performance with the automated bad channel detection (ABCD) algorithm for detecting bad channels.Approach.Six Neurologists, ten EEG Technologists, and six naïve research personnel (22 raters in total) were asked to rate 1440 real intracranial EEG channels as good or bad. Intra- and interrater kappa statistics were calculated for each group. We then compared each group to the ABCD algorithm which uses spectral and temporal domain features to classify channels as good or bad.Main results.Analysis of channel ratings from our participants revealed variable intra-rater reliability within each group, with no significant differences across groups. Inter-rater reliability was moderate among neurologists and EEG Technologists but minimal among naïve participants. Neurologists demonstrated a slightly higher consistency in ratings than EEG Technologists. Both groups occasionally misclassified flat channels as good, and participants generally focused on low-frequency content for their assessments. The ABCD algorithm, in contrast, relied more on high-frequency content. A logistic regression model showed a linear relationship between the algorithm's ratings and user responses for predominantly good channels, but less so for channels rated as bad. Sensitivity and specificity analyses further highlighted differences in rating patterns among the groups, with neurologists showing higher sensitivity and naïve personnel higher specificity.Significance.Our study reveals the bias in human assessments of intracranial electroencephalography (iEEG) data quality and the tendency of even experienced professionals to overlook certain bad channels, highlighting the need for standardized, unbiased methods. The ABCD algorithm, outperforming human raters, suggests the potential of automated solutions for more reliable iEEG interpretation and seizure characterization, offering a reliable approach free from human biases.

Timing matters for accurate identification of the epileptogenic zone.

OBJECTIVE: Interictal biomarkers of the epileptogenic zone (EZ) and their use in machine learning models open promising avenues for improvement of epilepsy surgery evaluation. Currently, most studies restrict their analysis to short segments of intracranial EEG (iEEG). METHODS: We used 2381 hours of iEEG data from 25 patients to systematically select 5-minute segments across various interictal conditions. Then, we tested machine learning models for EZ localization using iEEG features calculated within these individual segments or across them and evaluated the performance by the area under the precision-recall curve (PRAUC). RESULTS: On average, models achieved a score of 0.421 (the result of the chance classifier was 0.062). However, the PRAUC varied significantly across the segments (0.323-0.493). Overall, NREM sleep achieved the highest scores, with the best results of 0.493 in N2. When using data from all segments, the model performed significantly better than single segments, except NREM sleep segments. CONCLUSIONS: The model based on a short segment of iEEG recording can achieve similar results as a model based on prolonged recordings. The analyzed segment should, however, be carefully and systematically selected, preferably from NREM sleep. SIGNIFICANCE: Random selection of short iEEG segments may give rise to inaccurate localization of the EZ.

Optimizing automated detection of high frequency oscillations using visual markings does not improve SOZ localization.

OBJECTIVE: High frequency oscillations (HFOs) are a biomarker of the seizure onset zone (SOZ) and can be visually or automatically detected. In theory, one can optimize an automated algorithm's parameters to maximize SOZ localization accuracy; however, there is no consensus on whether or how this should be done. Therefore, we optimized an automated detector using visually identified HFOs and evaluated the impact on SOZ localization accuracy. METHODS: We detected HFOs in intracranial EEG from 20 patients with refractory epilepsy from two centers using (1) unoptimized automated detection, (2) visual identification, and (3) automated detection optimized to match visually detected HFOs. RESULTS: SOZ localization accuracy based on HFO rate was not significantly different between the three methods. Across patients, visually optimized detector settings varied, and no single set of settings produced universally accurate SOZ localization. Exploratory analysis suggests that, for many patients, detection settings exist that would improve SOZ localization. CONCLUSIONS: SOZ localization accuracy was similar for all three methods, was not improved by visually optimizing detector settings, and may benefit from patient-specific parameter optimization. SIGNIFICANCE: Visual HFO marking is laborious, and optimizing automated detection using visual markings does not improve localization accuracy. New patient-specific detector optimization methods are needed.