Worrell, Gregory A

Abstract Objective This study reports development and validation of a deep learning seizure detection algorithm for two‐channel subscalp electroencephalographic (EEG) recordings. Ultra‐long‐term monitoring of people with epilepsy may produce new insights into the timing and pattern of their seizures and may pave the way for novel therapeutic options. Although EEG is the accepted standard for epilepsy monitoring, ultra‐long‐term EEG recordings generate a massive quantity of data and are not reasonably reviewable in full by human readers. Methods The convolutional neural network‐bidirectional long short‐term memory (CNN‐BiLSTM) hybrid algorithm uses nine layers, operating on channel spectrograms 5 min in length with 50% overlap. Retrospective subscalp EEG data from 16 patients recorded at three centers were available for algorithm development and testing. EEG was recorded for a median of 63 days (range = 9–508), and a median of 17 seizures (range = 0–96) were recorded. Training data were augmented with scalp EEG seizures, and performance was benchmarked against a conventional spectral power classifier algorithm. We also evaluated an intrapatient training–testing approach where the algorithm was trained on the first 45% of data. Results The CNN‐BiLSTM trained on scalp EEG and subscalp EEG achieved an area under the receiver operating characteristic curve (AUROC) of .98 and an area under the precision–recall curve (AUPRC) of .50, which correspond to 94% sensitivity with 1.11 false detections per day. The same model trained on iEEG achieved only AUROC = .94 and AUPRC = .36. The conventional spectral band power detector achieved AUROC = .93 and AUPRC = .38. The CNN‐BiLSTM detector trained on the earliest half of the subscalp EEG data achieved AUROC = .93 and AUPRC = .37, corresponding to 87% sensitivity and 5.9 false detections per day. Significance High sensitivity and specificity are possible in automated seizure detection in two‐channel subscalp EEG data using a CNN‐BiLSTM framework. Performance of the detector is superior using subcutaneous EEG data for training rather than intracranial EEG, but addition of scalp EEG seizures for training was helpful.

AbstractObjectiveThe effects of brain stimulation for diseases like epilepsy are delayed, making stimulation optimization difficult. The parameters for anterior thalamic nuclei (ANT) deep brain stimulation (DBS) for focal drug‐resistant epilepsy management are often restricted to those used in the SANTE landmark trial. There is little evidence regarding effective alternatives, and low‐frequency stimulation is typically neglected. We prospectively compare a widely differing stimulation parameter set to typical settings.MethodsThis randomized, modified crossover, open trial compares the efficacy and safety of an alternative set of parameters using continuous low‐frequency stimulation (cLFS) with a longer pulse width (7 Hz, 200 μs, continuous) compared to SANTE's intermittent high‐frequency stimulation (iHFS) with a short pulse width (145 Hz, 90 μs, cycling 1 min on/5 min off). After 3 months on a randomly assigned first set, patients are switched to the other settings, unless seizure‐free. Patients are re‐evaluated after 3 more months, which marks the completion of the trial. After that, they can either remain on the same settings or switch back.ResultsSixteen patients with a median baseline seizure frequency of 13.8 seizures/month (interquartile range [IQR] = 2.7–22.8) were included in the analysis. At trial completion (median follow‐up = 30 weeks, IQR = 26–35), ANT‐DBS significantly reduced median seizure frequency (62%, IQR = 18–81, Wilcoxon test statistic [W] = 99, p = .008). Both iHFS (33%, IQR = 0–65, W = 81, p = .02) and cLFS (73%, IQR = 30–79, W = 105, p = .001) significantly reduced median seizure frequency. cLFS showed improved median seizure frequency reduction compared to iHFS (W = 63, p = .03) and was not associated with any moderate or severe adverse effects.SignificanceResults support cLFS for ANT‐DBS as a safe and effective alternative to typical iHFS parameters. Stimulation with widely differing parameter sets may be as effective as or, in some situations, more effective than typical stimulation parameters.

Abstract Electrical stimulation is increasingly used to modulate brain networks for clinical purposes. The basic unit of neurostimulation, a single electrical pulse, can travel through white matter to influence connected neuronal populations. However, the mechanisms by which it influences connected populations is not well understood: stimulation may excite, inhibit, or add noise to neuronal population activity. In this study, we investigated how single pulses modulate the neuronal processing of images in a well-controlled visual paradigm. In two human subjects implanted with iEEG electrodes for clinical purposes, single pulses were delivered to electrodes in white matter tracts connected to measurement electrodes in visual cortex. Images appeared on-screen at 0, 100, or 200 ms after each pulse. Using finite impulse response modeling, we decomposed the broadband and evoked potential responses into separate components induced by electrical stimulation and by visual processing. Single pulses induced transient broadband increases followed by suppression, but they did not modulate the visual broadband responses (i.e., stimulation response was additive to visual response). In contrast, single pulses elicited prominent brain stimulation evoked potentials and they modulated the visual evoked potentials. Specifically, visual evoked potentials were larger when stimulation occurred closer to visual onset. This indicates that a single electrical pulse can increase the strength or synchrony of visual inputs. Overall, these findings suggest that the effects of electrical stimulation in the visual system are two-fold: stimulation induces additive effects on broadband power, possibly by adding noise, and it interacts with synchronous visual inputs to amplify them.

Abstract Objective Noninvasive brain stimulation (NIBS) provides a treatment option for patients not eligible for surgical intervention or who seek low‐risk approaches and may be used in the hospital, clinic, and at home. Our objective is to summarize our single‐center experience with multiple NIBS approaches for the treatment of focal epilepsy. Methods A retrospective chart review identified drug‐resistant focal epilepsy patients who received NIBS as an epilepsy treatment at Mayo Clinic in Rochester, MN. Patients were typically treated as follows: (1) for TMS, 1 Hz stimulation was applied for five consecutive days in the neuromodulation clinic, (2) for outpatient tDCS, stimulation was applied for five consecutive days in the clinic, followed by optional treatment at home, and (3) for inpatient tDCS, stimulation was applied for three consecutive days. We analyzed continuous EEG data for the inpatient tDCS cohort and available HD‐EEG data for outpatient cohorts to quantify changes in interictal epileptiform discharges (IEDs) as a result of stimulation. Outcomes were assessed at 1 month for TMS and outpatient tDCS and 1 week for inpatient tDCS. Results Twenty‐four patients were treated with TMS ( n = 10) and tDCS ( n = 14, 9 as outpatients). The median age was 40 years (range 15–73). The median seizure reduction following stimulation was 50%. Fourteen patients (58%) were responders to treatment (TMS = 4/10, tDCS Outpatient = 7/9, tDCS Inpatient = 3/5). Five outpatient tDCS participants elected to continue treatment at home. Four TMS and four outpatient tDCS patients underwent high‐density EEG before and after 5 days of therapy. Following stimulation, IED rate was reduced in 4/5 inpatient tDCS patients, 4/4 outpatient tDCS patients, and 4/4 TMS patients. Two patients experienced an increase in seizure frequency (1 following TMS and 1 following outpatient tDCS), which returned to baseline 4–6 weeks after stimulation treatments were discontinued. Significance TMS and tDCS are potential treatment approaches for drug‐resistant focal epilepsy patients in the hospital, clinic, and home. They have a favorable safety profile and can lead to a reduction in IED rates and seizures. These results suggest further studies are needed to examine NIBS as a treatment for epilepsy. Plain Language Summary Noninvasive brain stimulation, such as transcranial magnetic stimulation and transcranial direct current stimulation, offers new treatment options for patients with focal seizures. This study reviewed the experience at Mayo Clinic using noninvasive brain stimulation in the hospital, clinic, and at‐home settings to treat seizures. The results showed an overall 50% median seizure reduction, and 58% of patients had at least a 50% reduction in seizures. Noninvasive brain stimulation is a promising treatment approach with a favorable safety profile.

AbstractObjectiveCognitive deficits are one of the most debilitating comorbidities in epilepsy and other neurodegenerative, neuropsychiatric, and neurodevelopmental brain disorders. Current diagnostic and therapeutic options are limited and lack objective measures of the underlying neural activities. In this study, electrophysiological biomarkers that reflect cognitive functions in clinically validated batteries were determined to aid diagnosis and treatment in specific brain regions.MethodsWe employed the Cambridge Neuropsychological Test Automated Battery (CANTAB) tasks to probe memory and executive functions in 86 patients with epilepsy undergoing clinical electroencephalography (EEG) monitoring. EEG electrode signals during performance of particular battery tasks were decomposed to identify specific frequency bands and cortical areas that differentiated patients with impaired, normal, and good standardized performance according to their age and gender.ResultsThe anterior prefrontal cortical EEG power in the theta frequency band was consistently lower in patients with impaired memory and executive function performance (z‐score < −1). This effect was evident in all four behavioral measures of executive, visual, spatial, and working memory functions and was confined to the cortical area of all four frontal pole electrodes (Nz, Fpz, Fp1, and Fp2).SignificanceTheta EEG power in the anterior prefrontal cortex provides simple, accessible, and objective electrophysiological measure of memory and executive functions in epilepsy. Our results suggest a feasible clinical biomarker for diagnosis, monitoring, and treatment of cognitive deficits with emerging targeted neuromodulation approaches.

Abstract Objective An accurate evaluation of behavioral responsiveness during and after seizures in people with epilepsy is critical for diagnosis and management. Current methods for assessing behavioral responsiveness are characterized by substantial variation, subjectivity, and limited reliability and reproducibility in ambulatory and epilepsy monitoring unit settings. In this study, we aimed to develop and implement a novel mobile platform for deployment of automated responsiveness testing in epilepsy—the ARTiE Watch—to facilitate standardized, objective assessments of behavioral responsiveness during and after seizures. Methods We prospectively recruited patients admitted to the epilepsy monitoring units for diagnostic evaluation and long‐term video‐electroencephalographic monitoring at Mayo Clinic and Yale New Haven Hospital. Participants wore the ARTiE Watch, a smartwatch paired with custom smartphone software integrated with cloud infrastructure allowing for remote activation of standardized assessment on the participants' smartwatches. The assessment consisted of 18 command prompts that test behavioral responsiveness across motor, language, and memory domains. Upon visually identifying an electrographic seizure during EMU monitoring, the BrainRISE platform was used to deploy the ARTiE Watch behavioral testing sequence. Responsiveness scoring was conducted on smartwatch files. Results Eighteen of 56 participants had a total of 39 electrographic seizures assessed with the ARTiE Watch. The 18 subjects with ARTiE Watch‐tested seizures had a total of 67 baseline (interictal) ARTiE Watch tests collected for analysis. The analysis showed distinct ARTiE Watch behavioral responsiveness phenotypes: (1) decreased responsiveness across all ARTiE Watch commands during seizure (ictal–postictal) periods compared (to baseline ( p < .0001), (2) decreased responsiveness in bilateral tonic–clonic seizures compared to baseline ( p < .0001) and compared to focal seizures ( p < .0001), and (3) decreased responsiveness during focal impaired awareness seizures compared to baseline ( p < .0001) and compared to focal aware seizures ( p < .001). Significance ARTiE Watch behavioral testing deployed utilizing a mobile cloud‐based platform is feasible and can provide standardized, objective behavioral responsiveness assessments during seizures.

AbstractObjectiveEpilepsy raises critical challenges to accurately localize the epileptogenic zone (EZ) to guide presurgical planning. Previous research has suggested that interictal spikes overlapping with high‐frequency oscillations, referred to here as pSpikes, serve as a reliable biomarker for EZ estimation, but there remains a question as to whether and to how pSpikes perform as compared to other types of epileptic spikes. This study aims to address this question by investigating the source imaging capabilities of pSpikes alongside other spike types.MethodsA total of 2819 interictal spikes from 76‐channel scalp electroencephalography (EEG) were analyzed in a cohort of 24 drug‐resistant focal epilepsy patients. All patients received surgical resection, and 16 were declared seizure‐free based on at least 1 year of postoperative follow‐up. A recently developed electrophysiological source imaging algorithm—fast spatiotemporal iteratively reweighted edge sparsity (FAST‐IRES)—was used for source imaging of the detected interictal spikes. The performance of 217 pSpikes was compared with 772 nSpikes (spikes with irregular high‐frequency activations), 1830 rSpikes (spikes with no high‐frequency activity), and all 2819 aSpikes (all interictal spikes).ResultsThe localization and extent estimation using pSpikes are concordant with the clinical ground truth; using pSpikes yields the best performance compared with nSpikes, rSpikes, and conventional spike imaging (aSpikes). For multiple spike type seizure‐free patients, the mean localization error for pSpike imaging was 6.8 mm, compared with 15.0 mm for aSpikes. The sensitivity, precision, and specificity were .41, .67, and .93 for pSpikes compared with .32, .48, and .93 for aSpikes.SignificanceThese results demonstrate the merits of noninvasive EEG source localization, and that (1) pSpike is a superior biomarker, outperforming conventional spike imaging for the localization of epileptic sources, and especially those with multiple irritative zones; and (2) FAST‐IRES provides accurate source estimation that is highly concordant with clinical ground truth, even in situations of single spike analysis with low signal‐to‐noise ratio.

Abstract There are limited treatment options for individuals with drug‐resistant idiopathic generalized epilepsy (IGE). Small, limited case series suggest that centromedian thalamus deep brain stimulation (CM‐DBS) may be an effective treatment option. The optimal CM‐DBS target for IGE is underexamined. Here, we present a retrospective analysis of CM‐DBS targeting and efficacy for five patients with drug‐resistant IGE. Volume of tissue activated (VTA) overlap with CM nucleus was performed using an open‐source toolbox. Median follow‐up time was 13 months. Median convulsive seizure frequency reduction was 66%. One patient had only absence seizures, with >99% reduction in absence seizure frequency. Four patients had electrode contacts positioned within the CM nucleus target, all of whom had >50% reduction in primary semiology seizure, with 85% median seizure reduction ( p = .004, paired‐sample t test). Volumetric “sweet‐spot” mapping revealed that best outcomes were correlated with stimulation of the middle ventral CM nucleus. Connectivity strength between the sweet‐spot region and central peri‐Rolandic cortex was increased significantly relative to other cortical regions ( p = 8.6 × 10 −4 , Mann–Whitney U test). Our findings indicate that CM‐DBS can be an effective treatment for patients with IGE, highlight the importance of accurate targeting and targeting analysis, and within the context of prior work, suggest that ideal CM‐DBS targets may be syndrome specific.