Park, Sihyeong

Abstract Objective Anterior nucleus of the thalamus (ANT) deep brain stimulation (DBS) is an effective therapy for frontotemporal and limbic drug‐resistant focal epilepsies. The pulvinar nucleus (PUL) is an emerging neuromodulation target for seizure networks involving the posterior quadrant and temporal lobe. This study evaluates the safety and effectiveness of combined bilateral ANT plus PUL DBS for posterior quadrant and temporal lobe epilepsies. Methods We conducted a single‐center retrospective cohort study of patients receiving combined ANT/PUL DBS between 2021 and 2025. Outcomes included median seizure rate reduction (MSR), responder rate (RR; >50% seizure frequency reduction), and stimulation‐related side effects, assessed across ANT, PUL, and concurrent ANT/PUL stimulation settings. Acute surgical complications were evaluated. Statistical significance was assessed using two‐sided Wilcoxon signed‐rank tests with Holm–Bonferroni correction. Results Nine patients with combined ANT/PUL DBS were identified, and eight patients had sufficient follow‐up for inclusion (median follow‐up = 20 months). Posterior predominant structural abnormalities were present in eight of nine patients. There were no acute surgical complications. Three instances of transient stimulation‐related side effects resolved with parameter adjustments, whereas one patient required rescue antiseizure medication in addition to programming changes. Eight patients received periods of ANT stimulation, six PUL stimulation, and seven combined ANT/PUL stimulation. Regarding overall outcome, at last follow‐up, there was a significant 80.8% MSR ( p = .0078) and 87.5% RR, with two patients receiving ANT and six ANT/PUL stimulation. Regarding target‐specific MSR, concurrent ANT/PUL DBS (68.8% MSR) showed a trend toward superior performance relative to ANT (43.3%) and PUL DBS (17.5%) alone, although subgroup analysis of target‐specific performance did not reach statistical significance. Significance Combined four‐lead ANT/PUL DBS appears safe and effective for patients with drug‐resistant posterior quadrant and temporal lobe epilepsies. Dual‐network neuromodulation may provide greater flexibility to individually optimize neuromodulation for complex seizure networks.

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.

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.