Localizing epileptogenic regions using high-frequency oscillations and machine learning.

Objective:Interictal high-frequency oscillations (HFOs) are a promising neurophysiological biomarker of the epileptogenic zone (EZ). However, objective criteria for distinguishing pathological from physiological HFOs remain elusive, hindering clinical application. We investigated whether the distinct mechanisms underlying pathological and physiological HFOs are encapsulated in their signal morphology in intracranial EEG (iEEG) recordings and whether this mechanism-driven distinction could be simulated by a deep generative model.Methods:In a retrospective cohort of 185 epilepsy patients who underwent iEEG monitoring, we analyzed 686,410 HFOs across 18,265 brain contacts. To learn morphological characteristics, each event was transformed into a time-frequency plot and input into a variational autoencoder. We characterized latent space clusters containing morphologically defined putative pathological HFOs (mpHFOs) using interpretability analysis, including latent space disentanglement and time-domain perturbation.Results:mpHFOs showed strong associations with expert-defined spikes and were predominantly located within the seizure onset zone (SOZ). Discovered novel pathological features included high power in the gamma (30–80 Hz) and ripple (>80 Hz) bands centered on the event. These characteristics were consistent across multiple variables, including institution, electrode type, and patient demographics. Predicting 12-month postoperative seizure outcomes using the resection ratio of mpHFOs outperformed unclassified HFOs (F1=0.72 vs. 0.68) and matched current clinical standards using SOZ resection (F1=0.74). Combining mpHFO data with demographic and SOZ resection status further improved prediction accuracy (F1=0.83).Interpretation:Our data-driven approach yielded a novel, explainable definition of pathological HFOs, which has the potential to further enhance the clinical use of HFOs for EZ delineation.

ObjectivePathological high-frequency oscillations (HFOs) in intracranial EEG are promising biomarkers of epileptogenic tissue, and their physiological counterparts play a role in sensorimotor and cognitive function. HFOs have also been found in scalp EEG, but an overview of all studies is lacking. In this systematic review, we assessed the methodology to detect scalp HFOs and their clinical potential. MethodsWe searched PubMed, Embase and the Cochrane Library for studies on HFOs in scalp EEG, and extracted methodological and clinical data. ResultsWe included 60 studies with data from 1149 unique individuals. Two-thirds of studies analyzed HFOs visually in the time or time–frequency domain, and one-third automatically with visual validation. Most studies evaluated interictal ripples during sleep in children. Pathological HFOs were overall better than spikes in localizing the epileptogenic zone and predicting outcome, correlated negatively with cognition and positively with disease activity and severity, and decreased after medical and surgical treatment. ConclusionsThe methodologies of the 60 studies were heterogeneous, but pathological scalp HFOs were clinically valuable as biomarkers in various situations, particularly in children with epilepsy. SignificanceThis systematic review gives an extensive overview of methodological and clinical data on scalp HFOs, establishing their clinical potential and discussing their limitations.

Objective. High-frequency oscillations (HFOs) are promising biomarkers for localizing epileptogenic brain tissue. Previous studies have revealed that HFOs that present concurrence with interictal epileptic discharges (IEDs) better delineate epileptogenic brain tissue, particularly for epilepsy patients with multitype interictal discharges. However, the analysis of noninvasively recorded epileptic HFOs involves many complex procedures, such as data preprocessing, detection and source localization, impeding the translation of this approach to clinical practice. Approach. To address these problems, we developed a graphical user interface (GUI)-based pipeline called EMHapp, which can be used for the automatic detection, source localization and visualization of HFO events concurring with IEDs in magnetoencephalography (MEG) signals by using a beamformer-based virtual sensor (VS) technique. An improved VS reconstruction method was developed to enhance the amplitudes of both HFO and IED VS signals. To test the capability of our pipeline, we collected MEG data from 11 complex focal epilepsy patients with surgical resections or seizure onset zones (SOZs) that were identified by intracranial electroencephalography. Main results. Our results showed that the HFO sources of eight patients were concordant with their resection margins or SOZs. Our proposed VS signal reconstruction approach achieved an 83.2% improvement regarding the number of detected HFO events and a 17.3% improvement in terms of the spatial overlaps between the HFO sources and the resection margins or SOZs in comparison with conventional VS reconstruction approaches. Significance. EMHapp is the first GUI-based pipeline for the analysis of epileptic magnetoencephalographic HFOs, which conveniently obtains HFO source locations using clinical data and enables direct translation to clinical applications.

Pathological high frequency oscillations associate with increased GABA synaptic activity in pediatric epilepsy surgery patients

In recent decades, a significant body of evidence based on invasive clinical research has showed that high-frequency oscillations (HFOs) are a promising biomarker for localization of the seizure onset zone (SOZ), and therefore, have the potential to improve postsurgical outcomes in patients with epilepsy. Emerging clinical literature has demonstrated that HFOs can be recorded noninvasively using methods such as scalp electroencephalography (EEG) and magnetoencephalography (MEG). Not only are HFOs considered to be a useful biomarker of the SOZ, they also have the potential to gauge disease severity, monitor treatment, and evaluate prognostic outcomes. In this article, we review recent clinical research on noninvasively detected HFOs in the human brain, with a focus on epilepsy. Noninvasively detected scalp HFOs have been investigated in various types of epilepsy. HFOs have also been studied noninvasively in other pathologic brain disorders, such as migraine and autism. Herein, we discuss the challenges reported in noninvasive HFO studies, including the scarcity of MEG and high-density EEG equipment in clinical settings, low signal-to-noise ratio, lack of clinically approved automated detection methods, and the difficulty in differentiating between physiologic and pathologic HFOs. Additional studies on noninvasive recording methods for HFOs are needed, especially prospective multicenter studies. Further research is fundamental, and extensive work is needed before HFOs can routinely be assessed in clinical settings; however, the future appears promising.

Epileptic spasms (ES) often become drug-resistant. To reveal the electrophysiological difference between children with ES (ES+) and without ES (ES-), we compared the occurrence rate (OR) of high-frequency oscillations (HFOs) and the modulation index (MI) of coupling between slow and fast oscillations. In ES+, we hypothesized that (1) pathological HFOs are more widely distributed and (2) slow oscillations show stronger coupling with pathological HFOs than in ES-. We retrospectively reviewed 24 children with drug-resistant multilobar onset epilepsy, who underwent intracranial video electroencephalography prior to multilobar resections. We measured the OR of HFOs and determined the electrodes with a high rate of HFOs by cluster analysis. We calculated MI, which reflects the degree of coupling between HFO (ripple/fast ripple [FR]) amplitude and 5 different frequency bands of delta and theta activities (0.5-1 Hz, 1-2 Hz, 2-3 Hz, 3-4 Hz, 4-8 Hz). In ES+ (n = 10), the OR(FRs) , the number of electrodes with high-rate FRs, and the MI(FRs & 3-4 Hz) in all electrodes were significantly higher than in ES- (n = 14). In both the ES+ and ES- groups, MI(ripples/FRs & 3-4 Hz) was the highest among the 5 frequency bands. Within the good seizure outcome group, the OR(FRs) and the MI(FRs & 3-4 Hz) in the resected area in ES+ were significantly higher than in ES- (OR[FRs] , P = .04; MI[FRs & 3-4 Hz] , P = .04). In ES+, the larger number of high-rate FR electrodes indicates more widespread epileptogenicity than in ES-. High values of OR(FRs) and MI(FRs & 3-4 Hz) in ES+ compared to ES- are a signature of the severity of epileptogenicity. We proved that ES+ children who achieved seizure freedom following multilobar resections exhibited strong coupling between slow oscillations and FRs.

Recent technological advances in signal acquisition and analysis have led to the discovery of high-frequency oscillatory events in animal models of mesial temporal lobe epilepsy (MTLE) and in the EEG of epileptic patients. These high-frequency oscillations (HFOs) in the 80–500 Hz frequency range are thought to reflect the activity of dysfunctional neuronal networks that may sustain epileptogenesis and ictogenesis. In this chapter, we review recent findings on HFOs in MTLE and their contribution to the identification of the pathophysiological mechanisms underlying the development of this epileptic condition. We describe the presumptive cellular and network mechanisms underlying ripples (80–200 Hz) and fast ripples (250–500 Hz) as well as their relation with interictal spikes and seizures. In addition, we discuss their contribution to epileptogenesis and ictogenesis in epileptic patients and animal models of MTLE.

Scalp EEG interictal high frequency oscillations as an objective biomarker of infantile spasms

High-frequency oscillations (HFOs) represent an electrographic biomarker of endogenous epileptogenicity and seizure-generating tissue that proved clinically useful in presurgical planning and delineating the resection area. In the neocortex, the clinical observations on HFOs are not sufficiently supported by experimental studies stemming from a lack of realistic neocortical epilepsy models that could provide an explanation of the pathophysiological substrates of neocortical HFOs. In this study, we explored pathological epileptiform network phenomena, particularly HFOs, in a highly realistic murine model of neocortical epilepsy due to focal cortical dysplasia (FCD) type II. FCD was induced in mice by the expression of the human pathogenic mTOR gene mutation during embryonic stages of brain development. Electrographic recordings from multiple cortical regions in freely moving animals with FCD and epilepsy demonstrated that the FCD lesion generates HFOs from all frequency ranges, i.e., gamma, ripples, and fast ripples up to 800 Hz. Gamma-ripples were recorded almost exclusively in FCD animals, while fast ripples occurred in controls as well, although at a lower rate. Gamma-ripple activity is particularly valuable for localizing the FCD lesion, surpassing the utility of fast ripples that were also observed in control animals, although at significantly lower rates. Propagating HFOs occurred outside the FCD, and the contralateral cortex also generated HFOs independently of the FCD, pointing to a wider FCD network dysfunction. Optogenetic activation of neurons carrying mTOR mutation and expressing Channelrhodopsin-2 evoked fast ripple oscillations that displayed spectral and morphological profiles analogous to spontaneous oscillations. This study brings experimental evidence that FCD type II generates pathological HFOs across all frequency bands and provides information about the spatiotemporal properties of each HFO subtype in FCD. The study shows that mutated neurons represent a functionally interconnected and active component of the FCD network, as they can induce interictal epileptiform phenomena and HFOs.

Real-time seizure onset detection has been an active area of research in the study of epilepsy. Electroencephalography (EEG) measurements are widely used in seizure detection due to their high temporal resolution, cost-effective, and noninvasive nature. Various approaches based on machine learning are used for epileptic seizure detection, but these approaches do not explicitly reveal the underlying dynamics, require larger datasets for training, and are computationally demanding. Although high-frequency oscillations (HFOs) are the new biomarkers of epilepsy, they cannot be used with the existing data acquisition systems as they require high sampling rates and high cutoff frequency of the used filters. In this article, we present a novel approach for real-time seizure detection using high spatial frequencies. Since eigenvalues of the graph Laplacian represent spatial frequencies, we conjecture that higher eigenvalues and eigenvectors will contain the detailed information of seizure and non-seizure brain states. Hence, we have formed sub-band characteristic response vector (sub-band CRV) using weighted sum of eigenvectors corresponding to higher spatial frequencies and analyzed it over time. We have used a publicly available dataset to demonstrate the efficacy of the proposed approach. We observed that the proposed approach performs satisfactorily well in real-time automated seizure detection without requiring any kind of prior training. Moreover, our approach is not only accurate in seizure detection but also is independent of sampling rates, hence can be implemented easily in clinical realm for developing an automated seizure detection tool with the existing data acquisition systems operating at low sampling rates.

Motivation: There is an ongoing search for definitive and reliable biomarkers to forecast or predict imminent seizure onset, but to date most research has been limited to EEG with sampling rates <1,000 Hz. High-frequency oscillations (HFOs) have gained acceptance as an indicator of epileptic tissue, but few have investigated the temporal properties of HFOs or their potential role as a predictor in seizure prediction. Here we evaluate time-varying trends in preictal HFO rates as a potential biomarker of seizure prediction.Methods: HFOs were identified for all interictal and preictal periods with a validated automated detector in 27 patients who underwent intracranial EEG monitoring. We used LASSO logistic regression with several features of the HFO rate to distinguish preictal from interictal periods in each individual. We then tested these models with held-out data and evaluated their performance with the area-under-the-curve (AUC) of their receiver-operating curve (ROC). Finally, we assessed the significance of these results using non-parametric statistical tests.Results: There was variability in the ability of HFOs to discern preictal from interictal states across our cohort. We identified a subset of 10 patients in whom the presence of the preictal state could be successfully predicted better than chance. For some of these individuals, average AUC in the held-out data reached higher than 0.80, which suggests that HFO rates can significantly differentiate preictal and interictal periods for certain patients.Significance: These findings show that temporal trends in HFO rate can predict the preictal state better than random chance in some individuals. Such promising results indicate that future prediction efforts would benefit from the inclusion of high-frequency information in their predictive models and technological architecture.

High-frequency oscillations (HFOs) in intracranial electroencephalograms of patients with epilepsy are regarded as promising biomarkers of epileptogenic zones. Their detection and classification can be achieved by visual assessment or automated approaches, although manual processing of large recordings can be laborious. As a result, an automated analysis scheme is indispensable to enable the clinical use of HFOs. In this paper, we present a two-stage strategy to detect and classify HFOs, which starts with a threshold-based approach to detect plausible HFO events followed by an event classification to discriminate different oscillations. Unlike existing approaches, the detection process in the proposed schemes starts by calculating various multi-channel features that allow interrelations among electrodes to be exploited for detection. On this basis, the detection thresholds are set epoch-by-epoch, relying on a two-component Gaussian mixture model to avoid threshold overestimation. The events deemed to be plausible HFOs are then subjected to classification. By simultaneously examining the raw data and time-frequency maps of these events, they are ultimately sorted into the following categories: HFOs, spikes, and spikes with HFOs, so that the oscillations solely caused by filtering sharp transients can be discriminated. Experimental results using simulated data and intracranial recordings from three epileptic patients demonstrate that our proposed schemes achieve promising sensitivity and precision, especially when the noise level is high.

The current study aimed to investigate the spatial and temporal patterns of high-frequency oscillations (HFOs) in the intra-/extrahippocampal areas during epileptogenesis. Local field potentials were bilaterally recorded from hippocampus (CA1), thalamus, motor cortex, and prefrontal cortex in 13 rats before and after intrahippocampal kainic acid (KA) lesions. HFOs in the ripple (100-200 Hz) and fast ripple (250-500 Hz) ranges were detected and their rates were computed during different time periods (1-5 weeks) after KA-induced status epilepticus (SE). Recurrent spontaneous seizures were observed in 7 rats after SE, and the other 6 rats did not develop epilepsy. During the latent period, the rate of hippocampal HFOs increased at the ipsilateral site of the KA lesion in both groups, and the HFO rate was significantly higher in the animals that later developed epilepsy. Animals that later developed epilepsy also demonstrated widespread appearance of HFOs, in both the ripple and the fast ripple range, whereas animals that did not develop epilepsy only exhibited changes in the ipsilateral intrahippocampal HFO rate. This study demonstrates an association between an increased rate of widespread HFOs and the later development of epilepsy, suggesting the formation of large-scale distributed pathological networks during epileptogenesis.

High-frequency oscillations (HFOs) are spontaneous electroencephalographic (EEG) events that occur within the frequency range of 80 to 500 Hz and consist of at least four distinct oscillations that stand out from the background activity. They can be further classified into "ripples" (80-250 Hz) and "fast ripples" (FR; 250-500 Hz) based on different frequency bands. Studies have indicated that HFOs may serve as important markers for identifying epileptogenic regions and networks in patients with refractory epilepsy. Furthermore, a higher extent of removal of brain regions generating HFOs could potentially lead to improved prognosis. However, the clinical application criteria for HFOs remain controversial, and the results from different research groups exhibit inconsistencies. Given this controversy, the aim of this study was to conduct a meta-analysis to explore the utility of HFOs in predicting postoperative seizure outcomes by examining the prognosis of refractory epilepsy patients with varying ratios of HFO removal. Prospective and retrospective studies that analyzed HFOs and postoperative seizure outcomes in epilepsy patients who underwent resective surgery were included in the meta-analysis. The patients in these studies were grouped based on the ratio of HFOs removed, resulting in four groups: completely removed FR (C-FR), completely removed ripples (C-Ripples), mostly removed FR (P-FR), and partial ripples removal (P-Ripples). The prognosis of patients within each group was compared to investigate the correlation between the ratio of HFO removal and patient prognosis. A total of nine studies were included in the meta-analysis. The prognosis of patients in the C-FR group was significantly better than that of patients with incomplete FR removal (odds ratio [OR] = 6.62; 95% confidence interval [CI]: 3.10-14.15; p < 0.00001). Similarly, patients in the C-Ripples group had a more favorable prognosis compared with those with incomplete ripples removal (OR = 4.45; 95% CI: 1.33-14.89; p = 0.02). Patients in the P-FR group had better prognosis than those with a majority of FR remaining untouched (OR = 6.23; 95% CI: 2.04-19.06; p = 0.001). In the P-Ripples group, the prognosis of patients with a majority of ripples removed was superior to that of patients with a majority of ripples remaining untouched (OR = 8.14; 95% CI: 2.62-25.33; p = 0.0003). There is a positive correlation between the greater removal of brain regions generating HFOs and more favorable postoperative seizure outcomes. However, further investigations, particularly through clinical trials, are necessary to justify the clinical application of HFOs in guiding epilepsy surgery.

High frequency oscillations (HFOs) are recognized as biomarkers for epileptogenic brain tissue. A remaining challenge for epilepsy surgery is the prospective classification of tissue sampled by individual electrode contacts. We analysed long-term invasive recordings of 20 consecutive patients who subsequently underwent epilepsy surgery. HFOs were defined prospectively by a previously validated, automated algorithm in the ripple (80–250 Hz) and the fast ripple (FR, 250–500 Hz) frequency band. Contacts with the highest rate of ripples co-occurring with FR over several five-minute time intervals designated the HFO area. The HFO area was fully included in the resected area in all 13 patients who achieved seizure freedom (specificity 100%) and in 3 patients where seizures reoccurred (negative predictive value 81%). The HFO area was only partially resected in 4 patients suffering from recurrent seizures (positive predictive value 100%, sensitivity 57%). Thus, the resection of the prospectively defined HFO area proved to be highly specific and reproducible in 13/13 patients with seizure freedom, while it may have improved the outcome in 4/7 patients with recurrent seizures. We thus validated the clinical relevance of the HFO area in the individual patient with an automated procedure. This is a prerequisite before HFOs can guide surgical treatment in multicentre studies.