2026年 · 第02卷第01期

The severity of traumatic brain injury (TBI) and its poor prognosis underscore the urgent need for more effective neuroprotective and neuroregenerative strategies. However, the pathological heterogeneity of TBI continues to hinder clinical therapeutic progress. Preclinical studies have shown that neural stem cells, through their regenerative and secretory properties, can suppress neuroinflammation and promote angiogenesis, synaptic regeneration, and neural circuit remodeling. On one hand, stem cell transplantation has entered clinical trials and demonstrates neuroprotective effects by promoting regeneration at the injury site. However, reliable evidence supporting its clinical benefits in severe TBI remains limited. Large-scale randomized controlled trials are needed to determine the optimal route, dosage, timing, and cell type for transplantation. On the other hand, emerging cell-free therapies and stem cell-scaffold combination strategies offer new directions for TBI treatment. This narrative review systematically explores recent advances, current developments, and ongoing challenges in stem cell therapy for TBI. It also highlights innovative approaches that integrate tissue engineering and genetic modification, which may drive transformative progress in TBI treatment through interdisciplinary collaboration.

Neurodegenerative disorders (NDDs) are increasingly being acknowledged as conditions characterized by network collapse rather than neuronal loss. Studies have shown that pathological proteins, such as β-amyloid (Aβ), tau, and α-synuclein, disseminate across structural and functional brain connections, utilizing highly interconnected hub areas characterized by increased metabolic demand. Progress in connectomics and network control theory indicates that hub susceptibility and modified network topology are fundamental to the development of Alzheimer’s disease (AD), Parkinson’s disease (PD), Huntington’s disease (HD), and amyotrophic lateral sclerosis (ALS). Moreover, multimodal biomarkers, including diffusion and functional magnetic resonance imaging (fMRI), positron emission tomography (PET), electroencephalography/magnetoencephalography (EEG/MEG), and digital phenotyping, facilitate the early identification of network dysfunction and allow classification into physiologically relevant subgroups. Electrophysiological assessments of oscillatory synchrony and complexity provide dynamic and economical indications of cortical hyperexcitability and network failure. The integration of digital and wearable technologies allows for an ongoing ecologically accurate assessment of cognitive and motor deterioration. Furthermore, therapeutic approaches are transitioning towards network-level restoration, encompassing neuromodulation (transcranial magnetic stimulation, deep brain stimulation, and closed-loop stimulation), polypharmacology, and adaptive cognitive or digital therapies aimed at reestablishing connections and plasticity. In conjunction with machine learning and in-silico modeling, individualized network "fingerprints" have been developed as a basis for connectomic precision medicine. This network-centric architecture allows for proactive, personalized, and potentially preventative strategies to decelerate or cease the progression of neurodegeneration.

Background: The accurate segmentation of meningiomas, the most common intracranial tumors in adults, in medical imaging data is an essential component of clinical workflows for diagnosis, treatment planning, and longitudinal monitoring. Manual segmentation is labor-intensive, subjective, and challenging for small, irregular, and atypical lesions. In recent years, deep learning has emerged as a transformative artificial intelligence (AI) tool that offers automated solutions for enhancing the efficiency, consistency, and scalability across diverse imaging settings. Methods: This review synthesizes findings from 34 peer-reviewed studies published between January 1, 2020, and October 31, 2025, identified using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, with a specific focus on AI-based methods for meningioma segmentation in magnetic resonance imaging (MRI) scans. We evaluate recent advances, including U-shaped convolutional neural network (U-Net) variants, attention-enhanced frameworks, and hybrid models. Additionally, we analyze the impact of dataset characteristics, imaging modalities, and pre-processing choices on performance. Results: The findings indicate that architectural innovation, rather than reliance on imaging protocols or pre-processing, is the primary driver of performance gains, with top models achieving Dice scores of up to 0.980 on large datasets. While numerous high-performing models rely on large public repositories, such as Figshare and brain tumor segmentation (BraTS) challenge, studies still employ custom datasets for targeted clinical use. Contrast-enhanced T1-weighted imaging is the most commonly used and effective imaging modality for meningioma segmentation. Nonetheless, challenges remain, including the segmentation of small tumors, generalizability across clinical sites, and real-time deployment of computationally demanding models. Conclusions: These insights highlight the need for future research to develop optimized architectures that generalize well across multi-institutional datasets while aligning with the computational constraints of real-world clinical environments.

Background: This systematic review and meta-analysis aimed to evaluate the safety and efficacy of stiripentol (STP) in patients with epilepsy, including those with Dravet syndrome (DS) and status epilepticus (SE). Methods: A comprehensive search of PubMed, Scopus, and Web of Science was conducted for articles published up to February 20, 2025. Eligible studies included clinical trials, cohort studies, case series, and cross-sectional analyses assessing STP in epilepsy. Two independent reviewers screened the studies and extracted data. Outcomes included ≥50 % seizure reduction, seizure freedom, and adverse events (AEs). Risk of bias was assessed using the Newcastle-Ottawa Scale and Joanna Briggs Institute tools. Meta-analyses were performed using a random-effects model to account for study heterogeneity. Results: Twenty-nine studies (n = 1445) were included. Overall, 52.2 % (95 % confidence interval [CI]: 43.1-61.1 %) of patients achieved ≥50 % seizure reduction, and 12.0 % (95 % CI: 7.7-18.1 %) achieved seizure freedom. Subgroup analyses showed ≥50 % seizure reduction in SE (73.9 %, 95 % CI: 28.6-95.2 %), generalized myoclonic seizures and/or generalized atypical absence seizures (65.0 %, 95 % CI: 50.7-77.0 %), focal epilepsy (56.5 %, 95 % CI: 48.3-64.3 %), DS (47.8 %, 95 % CI: 39.8-56.0 %), and generalized tonic-clonic seizures (46.2%, 95 % CI: 35.9-56.8 %). Seizure freedom was highest in the SE group (31.3 %, 95 % CI: 16.2-51.6 %) and lowest in the DS group (7.9 %, 95 % CI: 4.6-13.1 %). The most common AEs were drowsiness (28.2 %, 95 % CI: 18.6-37.7 %), behavioral changes (24.5 %, 95 % CI: 9.2-39.7 %), weight loss (22.6 %, 95 % CI: 14.2-31.1 %), and decreased appetite (22.0 %, 95 % CI: 14.7-29.3 %). Treatment discontinuation due to AEs occurred in 7.2 % (95 % CI: 3.6-10.8 %) of patients. Conclusions: STP is a well-tolerated and effective adjunctive treatment for refractory epilepsy, particularly in SE and DS. It significantly reduces seizure frequency, with a meaningful rate of seizure freedom in SE. Most AEs were mild, supporting the clinical value of STP in patients with difficult-to-treat epilepsy.

Background: Timely and accurate prediction of clinical outcomes in patients with cardio-cerebral infarction (CCI) is essential for optimizing treatment strategies and improving prognosis. This study aimed to develop an interpretable machine learning model for predicting all-cause in-hospital mortality in patients with CCI. Methods: A total of 4105 patients with CCI from a multicenter database were retrospectively analyzed. Nine machine learning (ML) models were constructed to predict in-hospital mortality. Feature selection was performed using the recursive feature elimination algorithm. Model performance was evaluated using the F1 score, area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy. The model with the highest predictive performance was further interpreted using the SHapley Additive exPlanations (SHAP) method to provide individualized explanations for predictions. Results: Of the 4105 patients, 2071 (50.5 %) were female, with a median age of 79 years (interquartile range [IQR]: 72, 85). After feature selection, eight of the initial 43 clinical variables were retained for model development. Among the nine models, the Random Forest (RF) model achieved the highest predictive performance, with an AUC of 0.88 in the test cohort. Feature importance analysis identified serum potassium (K), N-terminal pro-B-type natriuretic peptide (NT-proBNP), troponin T, cholinesterase, creatinine, age, albumin (ALB), and platelet count (PLT) as the most influential predictors. SHAP visualization was applied to enhance the interpretability of the RF model. Conclusions: The proposed ML-based risk prediction model demonstrated robust performance in predicting in-hospital mortality among patients with CCI. The integration of SHAP analysis provided transparent, patient-specific risk interpretation, thereby supporting clinical understanding of critical prognostic factors and enabling personalized decision-making.

Background: Essential tremor (ET) with soft neurological signs, such as questionable dystonia, was defined as ET plus by the Movement Disorder Society (MDS). Ventral intermediate nucleus (Vim)-posterior subthalamic area (PSA) double-target deep brain stimulation (DBS) is a novel method of inhibiting tremors in patients with ET. Despite the established efficacy of the Vim and PSA as individual DBS targets in ET, their combined application in ET plus dystonia remains understudied. This study aimed to investigate the efficacy of Vim-PSA double-target DBS in patients with ET plus dystonia, and explore the potential mechanisms of DBS in ET plus dystonia. Methods: We reviewed 13 patients with ET plus dystonia who underwent bilateral Vim-PSA DBS between March 2022 and January 2025. Two patients were excluded because of their high BFMDRS scores, which indicated overt rather than "questionable" dystonia. The severity of tremor and dystonia was assessed using the Fahn-Tolosa-Marin Tremor Rating Scale (FTMTRS) and Burke-Fahn-Marsden Dystonia Rating Scale (BFMDRS), respectively. Clinical outcomes were evaluated by comparing optimal follow-up stimulation-on state with the baseline values. Electrodes, nuclei, and fiber tracts were reconstructed using Lead-DBS software. Results: All patients underwent bilateral interleaving stimulation (ILS) mode DBS to ensure optimal efficacy. Following high-frequency stimulation (150-185 Hz) of the Vim and PSA, FTMTRS and BFMDRS scores declined by 82.07 % and 73.92 %, respectively. The anterior-ventral margin of the Vim and the realm of the fasciculus cerebellothalamicus (fct) in the PSA were effective stimulation regions. Gait and postural disorders were common adverse effects. Conclusions: Our findings suggest that Vim-PSA double-targeting DBS is a potential therapeutic approach for ET plus dystonia. High-frequency bilateral ILS may be an effective stimulation pattern for alleviating tremors and dystonia.