Machine learning approach to detect focal-onset seizures in the human anterior nucleus of the thalamus

Epileptiform discharges in the anterior thalamus of epilepsy patients

Anterior thalamus (ANT) deep-brain stimulation (DBS) is an approved therapy for drug resistant epilepsy. We aimed to identify interictal epileptiform discharges (IED) in the ANT and to investigate their relationship with surface IEDs. Fifteen patients were monitored for two consecutive nights with externalized thalamic leads to analyze the intrathalamic epileptiform activities (TIED). Forty-six % of all contacts were located within the ANT. We found that all the responders had TIEDs within the ANT, while this held true only for 44% of the non-responders. The overall response rate (RR) at 1-year follow-up was 40%, while it was 44% in bilateral ANT hit patients and 45% in epileptic focus side hit. However, in case of TIEDs present in the focus side the RR reached as high as 71%. TIED activity may prove the pathophysiological connection to the seizure focus, and stimulation of this area might have a better suppressing effect on seizures.

Sensing with deep brain stimulation device in epilepsy: Aperiodic changes in thalamic local field potential during seizures

OBJECTIVE

Thalamic deep brain stimulation (DBS) is an effective therapeutic option in patients with drug-resistant epilepsy. Recent DBS devices with sensing capabilities enable chronic, outpatient local field potential (LFP) recordings. Whereas beta oscillations have been demonstrated to be a useful biomarker in movement disorders, the clinical utility of DBS sensing in epilepsy remains unclear. Our aim was to determine LFP features that distinguish ictal from inter-ictal states, which may aid in tracking seizure outcomes with DBS.

METHODS

Electrophysiology data were obtained from DBS devices implanted in the anterior nucleus (N = 12) or centromedian nucleus (N = 2) of the thalamus. Power spectra recorded during patient/caregiver-marked seizure events were analyzed with a method that quantitatively separates the oscillatory and non-oscillatory/aperiodic components of the LFP using non-parametric statistics, without the need for pre-specification of the frequency bands of interest. Features of the LFP parameterized using this algorithm were compared with those from inter-ictal power spectra recorded in clinic.

RESULTS

Oscillatory activity in multiple canonical frequency bands was identified from the power spectra in 86.48% of patient-marked seizure events. Delta oscillations were present in all patients, followed by theta (N = 10) and beta (N = 9). Although there were no differences in oscillatory LFP features between the ictal and inter-ictal states, there was a steeper decline in the 1/f slope of the aperiodic component of the LFP during seizures.

SIGNIFICANCE

Our work highlights the potential and shortcomings of chronic LFP recordings in thalamic DBS for epilepsy. Findings suggest that no single frequency band in isolation clearly differentiates seizures, and that features of aperiodic LFP activity may be clinically-relevant biomarkers of seizures.

Neuromodulation of the anterior thalamus: Current approaches and opportunities for the future

The role of thalamocortical circuits in memory has driven a recent burst of scholarship, especially in animal models. Investigating this circuitry in humans is more challenging. And yet, the development of new recording and stimulation technologies deployed for clinical indications has created novel opportunities for data collection to elucidate the cognitive roles of thalamic structures. These technologies include stereoelectroencephalography (SEEG), deep brain stimulation (DBS), and responsive neurostimulation (RNS), all of which have been applied to memory-related thalamic regions, specifically for seizure localization and treatment. This review seeks to summarize the existing applications of neuromodulation of the anterior thalamic nuclei (ANT) and highlight several devices and their capabilities that can allow cognitive researchers to design experiments to assay its functionality. Our goal is to introduce to investigators, who may not be familiar with these clinical devices, the capabilities, and limitations of these tools for understanding the neurophysiology of the ANT as it pertains to memory and other behaviors. We also briefly cover the targeting of other thalamic regions including the centromedian (CM) nucleus, dorsomedial (DM) nucleus, and pulvinar, with associated potential avenues of experimentation.

Deep brain stimulation of thalamus for epilepsy

Neuromodulation (neurostimulation) is a relatively new and rapidly growing treatment for refractory epilepsy. Three varieties are approved in the US: vagus nerve stimulation (VNS), deep brain stimulation (DBS) and responsive neurostimulation (RNS). This article reviews thalamic DBS for epilepsy. Among many thalamic sub-nuclei, DBS for epilepsy has been targeted to the anterior nucleus (ANT), centromedian nucleus (CM), dorsomedial nucleus (DM) and pulvinar (PULV). Only ANT is FDA-approved, based upon a controlled clinical trial. Bilateral stimulation of ANT reduced seizures by 40.5% at three months in the controlled phase (p = .038) and 75% by 5 years in the uncontrolled phase. Side effects related to paresthesias, acute hemorrhage, infection, occasional increased seizures, and usually transient effects on mood and memory. Efficacy was best documented for focal onset seizures in temporal or frontal lobe. CM stimulation may be useful for generalized or multifocal seizures and PULV for posterior limbic seizures. Mechanisms of DBS for epilepsy are largely unknown, but animal work points to changes in receptors, channels, neurotransmitters, synapses, network connectivity and neurogenesis. Personalization of therapies, in terms of connectivity of the seizure onset zone to the thalamic sub- nucleus and individual characteristics of the seizures, might lead to improved efficacy. Many questions remain about DBS, including the best candidates for different types of neuromodulation, the best targets, the best stimulation parameters, how to minimize side effects and how to deliver current noninvasively. Despite the questions, neuromodulation provides useful new opportunities to treat people with refractory seizures not responding to medicines and not amenable to resective surgery.

A Feature Extraction Method for Seizure Detection Based on Multi-Site Synchronous Changes and Edge Detection Algorithm

Automatic detection of epileptic seizures is important in epilepsy control and treatment, and specific feature extraction assists in accurate detection. We developed a feature extraction method for seizure detection based on multi-site synchronous changes and an edge detection algorithm. We investigated five chronic temporal lobe epilepsy rats with 8- and 12-channel detection sites in the hippocampus and limbic system. Multi-site synchronous changes were selected as a specific feature and implemented as a seizure detection method. For preprocessing, we used magnitude-squared coherence maps and Canny edge detection algorithm to find the frequency band with the most significant change in synchronization and the important channel pairs. In detection, we used the maximal cross-correlation coefficient as an indicator of synchronization and the correlation coefficient curves' average value and standard deviation as two detection features. The method achieved high performance, with an average 96.60% detection rate, 2.63/h false alarm rate, and 1.25 s detection delay. The experimental results show that synchronization is an appropriate feature for seizure detection. The magnitude-squared coherence map can assist in selecting a specific frequency band and channel pairs to enhance the detection result. We found that individuals have a specific frequency band that reflects the most significant synchronization changes, and our method can individually adjust parameters and has good detection performance.

Thalamic stereoelectroencephalography in epilepsy surgery: a scoping literature review

OBJECTIVE

Stereoelectroencephalography (sEEG) is a well-established surgical method for defining the epileptogenic network. Traditionally reserved for identifying discrete cortical regions for resection or ablation, sEEG in current practice is also used for identifying more broadly involved subcortical epileptic network components, driven by the availability of brain-based neuromodulation strategies. In particular, sEEG investigations including thalamic nuclei are becoming more frequent in parallel with the increase in therapeutic strategies involving thalamic targets such as deep brain stimulation (DBS) and responsive neurostimulation (RNS). The objective to this study was to evaluate existing evidence and trends regarding the purpose, techniques, and relevant electrographic findings of thalamic sEEG.

METHODS

MEDLINE and Embase databases were systematically queried for eligible peer-reviewed studies involving sEEG electrode implantation into thalamic nuclei of patients with epilepsy. Available data were abstracted concerning preoperative workup and purpose for implanting the thalamus, thalamic targets and trajectories, and electrophysiological methodology and findings.

RESULTS

sEEG investigations have included thalamic targets for both basic and clinical research purposes. Medial pulvinar, dorsomedial, anterior, and centromedian nuclei have been the most frequently studied. Few studies have reported any complications with thalamic sEEG implantation, and no studies have reported long-term complications. Various methods have been utilized to characterize thalamic activity in epileptic disorders including evoked potentials, power spectrograms, synchronization indices, and the epileptogenicity index. Thalamic intracranial recordings are beginning to be used to guide neuromodulation strategies including RNS and DBS, as well as to understand complex, network-dependent seizure disorders.

CONCLUSIONS

Inclusion of thalamic coverage during sEEG evaluation in drug-resistant epilepsy is a growing practice and is amenable to various methods of electrographic data analysis. Further study is required to establish well-defined criteria for thalamic implantation during invasive investigations as well as safety and ethical considerations.

Time-Series Generative Adversarial Network Approach of Deep Learning Improves Seizure Detection From the Human Thalamic SEEG

Seizure detection algorithms are often optimized to detect seizures from the epileptogenic cortex. However, in non-localizable epilepsies, the thalamus is frequently targeted for neuromodulation. Developing a reliable seizure detection algorithm from thalamic SEEG may facilitate the translation of closed-loop neuromodulation. Deep learning algorithms promise reliable seizure detectors, but the major impediment is the lack of larger samples of curated ictal thalamic SEEG needed for training classifiers. We aimed to investigate if synthetic data generated by temporal Generative Adversarial Networks (TGAN) can inflate the sample size to improve the performance of a deep learning classifier of ictal and interictal states from limited samples of thalamic SEEG. Thalamic SEEG from 13 patients (84 seizures) was obtained during stereo EEG evaluation for epilepsy surgery. Overall, TGAN generated synthetic data augmented the performance of the bidirectional Long-Short Term Memory (BiLSTM) performance in classifying thalamic ictal and baseline states. Adding synthetic data improved the accuracy of the detection model by 18.5%. Importantly, this approach can be applied to classify electrographic seizure onset patterns or develop patient-specific seizure detectors from implanted neuromodulation devices.

Chronic electroencephalography in epilepsy with a responsive neurostimulation device: current status and future prospects

INTRODUCTION

Implanted neurostimulation devices are gaining traction as therapeutic options for people with certain forms of drug-resistant focal epilepsy. Some of these devices enable chronic electroencephalography (cEEG), which offers views of the dynamics of brain activity in epilepsy over unprecedented time horizons.

AREAS COVERED

This review focuses on clinical insights and basic neuroscience discoveries enabled by analyses of cEEG from an exemplar device, the NeuroPace RNS® System. Applications of RNS cEEG covered here include counting and lateralizing seizures, quantifying medication response, characterizing spells, forecasting seizures, and exploring mechanisms of cognition. Limitations of the RNS System are discussed in the context of next-generation devices in development.

EXPERT OPINION

The wide temporal lens of cEEG helps capture the dynamism of epilepsy, revealing phenomena that cannot be appreciated with short duration recordings. The RNS System is a vanguard device whose diagnostic utility rivals its therapeutic benefits, but emerging minimally invasive devices, including those with subscalp recording electrodes, promise to be more applicable within a broad population of people with epilepsy. Epileptology is on the precipice of a paradigm shift in which cEEG is a standard part of diagnostic evaluations and clinical management is predicated on quantitative observations integrated over long timescales.