Neuroscience informatics最新文献

{"title":"From text to code – Leveraging machine learning for neurology outpatient clinical coding","authors":"Elena Purcaru ,&nbsp;Michael George ,&nbsp;Matthew Stammers ,&nbsp;Christopher Kipps","doi":"10.1016/j.neuri.2026.100257","DOIUrl":"10.1016/j.neuri.2026.100257","url":null,"abstract":"<div><h3>Background</h3><div>Most neurological care is delivered in outpatient settings without mandated clinical coding. The clinical records remain stored as unstructured text with inconsistent formatting. There is a significant opportunity to increase the value of these data through automated clinical coding utilising natural language processing (NLP). While existing models for full ICD-10 clinical coding lack sufficient accuracy for clinical use, 60 % of neurology outpatient cases fall into just five diagnostic categories. This suggests that a simplified coding system could enhance feasibility and serve as a foundation for more complex coding schemes.</div></div><div><h3>Objective</h3><div>We propose a simplified coding system of 29 codes for neurology outpatient episodes. We evaluate several machine learning methods in a supervised single-label classification task on real-world outpatient care notes.</div></div><div><h3>Methods</h3><div>We collected outpatient care notes created between 15 November 2018 and 2 December 2022. The training dataset included 14,917 care notes, most of which were annotated with ICD-10 codes during routine care and subsequently mapped to 29 simplified diagnostic categories. An external validation set of 1,042 randomly selected encounters was retrospectively coded.</div><div>Models included logistic regression, support vector machine, bidirectional LSTM, BERT-based models (DistilBERT, RoBERTa), and a generative large language model (LLM), Mistral 7B. All but the LLM were trained via 10-fold stratified cross-validation; final models were trained on the complete dataset.</div></div><div><h3>Results</h3><div>DistilBERT and RoBERTa outperformed traditional models, with F1-scores of 81.73 (95 % CI: 79.02–84.13) and 81.16 (95 % CI: 78.84–83.76), respectively. The LLM–DistilBERT hybrid performed worse than all but BiLSTM and produced “medical hallucinations,” making it unsuitable for clinical use. The training data were highly imbalanced. BERT-based models showed strong performance on high-frequency categories, with F1-scores over 85 % for the top five classes. At a 0.85 confidence threshold, DistilBERT achieved 96 % accuracy on 64 % of the external validation set.</div></div><div><h3>Conclusions</h3><div>BERT-based NLP models perform well in classifying neurology outpatient clinic notes when a reduced set of diagnostic categories is used. In a human-in-the-loop workflow, such models can meaningfully reduce the manual coding workload while preserving accuracy. To our knowledge, this is the first applied study of automated clinical coding in neurology outpatient care.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100257"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146077248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advances in acquisition and post-processing optimization of IVIM MRI for brain imaging: A systematic review","authors":"Abhijith S. ,&nbsp;Saikiran Pendem ,&nbsp;Rajagopal Kadavigere ,&nbsp;Priyanka ,&nbsp;Dharmesh Singh ,&nbsp;Priya P.S.","doi":"10.1016/j.neuri.2025.100256","DOIUrl":"10.1016/j.neuri.2025.100256","url":null,"abstract":"<div><h3>Purpose</h3><div>Diffusion-weighted MRI is widely used to probe brain microstructure, but its signal reflects both diffusion and perfusion effects. Intravoxel Incoherent Motion (IVIM) MRI enables non-contrast separation of these components, offering potential clinical value in neuroimaging. However, clinical translation remains limited due to variability in acquisition and post-processing methods. This systematic review evaluates optimization strategies aimed at improving the accuracy, reproducibility, and clinical utility of IVIM parameters in brain.</div></div><div><h3>Methods</h3><div>Registered in PROSPERO and conducted according to PRISMA guidelines, a systematic search across five databases was performed. Original peer-reviewed studies focusing on optimization of IVIM acquisition or post-processing in human brain imaging were included, while reviews and studies lacking methodological detail were excluded. Study quality was assessed using a customized QUADAS-2 tool. Due to methodological heterogeneity, an effect direction plot was applied instead of meta-analysis.</div></div><div><h3>Results</h3><div>Out of 1,668 identified records, 14 studies were included. Acquisition strategies such as optimised b-value sampling, cardiac gating, and advanced sequences reduced parameter variability by up to 40 %. Post-processing methods, including Bayesian fitting, deep learning–based models, and advanced denoising, improved parameter accuracy by up to 99 % and precision by up to 95 %. Effect direction analysis demonstrated significant positive effects on accuracy and clinical utility (p &lt; 0.001) and repeatability (p &lt; 0.05), while scan-time reduction showed no significant benefit (p &gt; 0.05). No study reported gold-standard validation.</div></div><div><h3>Conclusion</h3><div>Optimization of IVIM acquisition and post-processing enhances parameter robustness and reproducibility in brain MRI, though protocol heterogeneity remains a major obstacle to standardization and clinical adoption.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100256"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924932","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"NeuroFusion: A forensic enriched ensemble framework for cerebellum disease classification","authors":"Abu Hanzala ,&nbsp;Md Sajjad ,&nbsp;Tanjila Akter ,&nbsp;Harpreet Kaur ,&nbsp;Md Sadekur Rahman","doi":"10.1016/j.neuri.2025.100251","DOIUrl":"10.1016/j.neuri.2025.100251","url":null,"abstract":"<div><div>Accurate and timely classification of cerebellar diseases is crucial for effective diagnostic, yet it remains challenging due to the inherent heterogeneity of these disorders and the subtlety of their neuroimaging manifestations. This study investigated a novel multi-stage ensemble framework integrating SE blocks and segmentation-assisted augmentation tailored for limited cerebellum disease MRI data. Dataset included 3296 MRI scans from four classes and we divided dataset into three parts: training, testing, and validation, and their ratio was 64:20:16. However, we performed image forensic analysis on it, such as Error Level Analysis (ELA) and Noise Residual Analysis (NRA). This study used deep learning architectures that can automatically classify cerebellum diseases and compared these models, which included six D-CNNs models, six transfer learning models, and three ensemble models. Another important contribution of our study is the significant improvement in the classification efficiency by strategically integrating squeeze and excitation and label smoothing techniques. We show that fine-tuning significantly improves the diagnostic accuracy of both D-CNNs and transfer learning models on cerebellum MRI data. Notably, our combined models consistently achieve higher performance, with FusionNet-6 reaching an exceptional accuracy of 99.83 %. K-fold cross-validation was performed, yielding consistently high performance with per-class sensitivity and specificity above 99 %. The study also greatly enhances the impact of dataset augmentation techniques, including the use of segmented data to reveal complex interactions that can enhance the performance of some models or, in some cases, dramatically reduce the performance of specific models. These results underscore the immense potential of deep learning ensembles to provide highly accurate and robust diagnostic support for cerebellum diseases, paving the way for more objective and efficient clinical workflows.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100251"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Age-related changes in brain fiber pathways based on directional decomposition of DTI tractograms","authors":"My N. Nguyen,&nbsp;Yoshiki Kubota,&nbsp;Akimasa Hirata","doi":"10.1016/j.neuri.2025.100255","DOIUrl":"10.1016/j.neuri.2025.100255","url":null,"abstract":"<div><div>This study investigated age-related changes of brain fiber pathways from diffusion tensor imaging (DTI) tractograms with directional decomposition. Two hundred subjects were stratified into three age groups. Tractograms were generated at two levels: from individual DTI images (subject-level), and from group-averaged images (group-level). Fiber tracking was performed within the cerebral white matter, brainstem, thalamus, and cerebellum at both the levels. Each tractogram was decomposed into directional tracts. At the subject-level, original and decomposed tracts were used to quantify tract density and correlations with age. Tract density was highest in the thalamus and brainstem, while the cerebellum showed the greatest inter-subject variability. Tract count exhibited some significant correlations with age: in cerebral white matter, it decreased overall, especially along S-I and A-P directions; in thalamus, S-I and A-P tracts decreased, while L-R and mixed-direction tracts increased. The brainstem tracts demonstrated its overall stability during aging. At the group level, ∼60 % of brainstem tracts were oriented along the S–I direction, and ∼64 % of cerebellar tracts along the A–P direction. Notably, the posterolateral tracts of the cerebellum showed asymmetry, with the left side associated with visuospatial processing, containing fewer tracts than the right side associated with language pathways. These findings highlight region- and direction-specific changes with age, revealing structural patterns that are not captured by conventional scalar measures. They suggested candidate biomarkers for brain aging and provided useful references for longitudinal neuroimaging and brain stimulation studies, with potential applications in the early detection of neurodegeneration and optimization of stimulation strategies.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100255"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924928","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Attention-Gated CNN and discrete wavelet transform based ensemble framework for brain hemorrhage classification","authors":"Srutanik Bhaduri ,&nbsp;Rasel Mondal ,&nbsp;Prateek Sarangi ,&nbsp;Vinod Kumar Kurmi ,&nbsp;Swati Goyal ,&nbsp;Lovely Kaushal ,&nbsp;Mahek Sodani ,&nbsp;Tanmay Basu","doi":"10.1016/j.neuri.2025.100243","DOIUrl":"10.1016/j.neuri.2025.100243","url":null,"abstract":"<div><div>Brain hemorrhage, or Intracranial Hemorrhage (ICH), is a critical medical condition requiring rapid diagnosis. Existing Convolutional Neural Network (CNN) models often struggle to differentiate similar hemorrhage subtypes like Epidural (EDH) and Subdural (SDH) due to a lack of specific spatial feature identification. This study aims to develop a robust classification framework to address this challenge. We propose an ensemble framework combining two complementary models. The first is an attention-gated 2D CNN designed to highlight subtle hemorrhagic regions. The second is a multi-level Discrete Wavelet Transform (DWT) model that analyzes images in the frequency domain to capture deeper contextual and textural information from the 3D brain volume. The proposed ensemble model was evaluated on the RSNA, CQ500, and a new GMC clinical dataset. The empirical study demonstrates that our model consistently outperforms state-of-the-art methods across standard evaluation metrics, including accuracy, macro-averaged AUC-ROC, specificity, sensitivity, and F1-score. The novel ensembling of an attention-gated CNN and a DWT-based model provides a more comprehensive feature representation, leading to significantly improved accuracy and robustness in ICH classification, particularly in distinguishing challenging subtypes like EDH and SDH.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100243"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145580157","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancing brain tumor diagnostics and therapy (2023–2025): A global bibliometric perspective on innovation and collaboration","authors":"Rosnah Sutan ,&nbsp;Afzal Hussain ,&nbsp;Rizal Abdul Manaf ,&nbsp;Zaleha Md Isa ,&nbsp;Ashfaq Hussain","doi":"10.1016/j.neuri.2025.100250","DOIUrl":"10.1016/j.neuri.2025.100250","url":null,"abstract":"<div><h3>Background</h3><div>Brain tumors present a daunting clinical challenge, necessitating unwavering innovation in diagnostics, imaging, and therapeutics. Emerging advances in artificial intelligence (AI), molecular biomarkers, and neuroimaging have transformed the research landscape.</div></div><div><h3>Objective</h3><div>The current research conducted bibliometric analysis to map global research trends in brain tumor diagnosis, imaging, and treatment strategies between 2023 and 2025, with a particular focus on AI applicability and biomarker-driven precision medicine.</div></div><div><h3>Methods</h3><div>A systematic literature search of the Scopus database was performed in July 2025 for English original research articles between 2023 and 2025. The search keywords included: “brain tumor,” “glioma,” “glioblastoma,” “meningioma,” “astrocytoma,” “diagnosis,” “MRI,” “CT,” “artificial intelligence,” “deep learning,” “machine learning,” “radiotherapy,” “chemotherapy,” “surgery,” “biomarkers,” “prognosis,” “segmentation,” and “classification.” Bibliographic data were analyzed using Biblioshiny to explore publication output, citation impact, prominent authors, institutional productivity, keyword trends, and collaboration networks.</div></div><div><h3>Results</h3><div>The analysis included 23,496 papers from over 93,000 researchers. It indicated a research boom in AI-enhanced diagnostics, radiomics, and individualized treatments. Both China and the U.S. were leading producers, but the U.S. recorded greater international collaboration and citation impact. Glioma classification, MRI-based segmentation, and deep learning applications were the most common topics. Collaboration networks were geographically focused, with a particular concentration in East Asia.</div></div><div><h3>Conclusion</h3><div>Brain tumor research is rapidly moving towards precision and AI-driven strategies. While there is a growing scientific output, more international and intersectoral collaboration is needed to make these advances translate to equitable clinical gain.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100250"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924929","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TRELLIS -enhanced surface features for comprehensive intracranial aneurysm analysis","authors":"Clément Hervé,&nbsp;Paul Garnier,&nbsp;Jonathan Viquerat,&nbsp;Elie Hachem","doi":"10.1016/j.neuri.2026.100259","DOIUrl":"10.1016/j.neuri.2026.100259","url":null,"abstract":"<div><div>Intracranial aneurysms pose a significant clinical risk yet are difficult to detect, delineate, and model due to limited annotated 3D data. We propose a cross-domain feature-transfer approach that leverages the latent geometric embeddings learned by TRELLIS, a generative model trained on large-scale non-medical 3D datasets, to augment neural networks for aneurysm analysis. By replacing conventional point normals or mesh descriptors with TRELLIS surface features, we systematically enhance three downstream tasks: (i) classifying aneurysms versus healthy vessels in the Intra3D dataset, (ii) segmenting aneurysm and vessel regions on 3D meshes, and (iii) predicting time-evolving blood-flow fields using a graph neural network on the AnXplore dataset. Our experiments show that the inclusion of these features yields strong gains in accuracy, F1-score, and segmentation quality over state-of-the-art baselines, and reduces simulation error by 15%. These results illustrate the broader potential of transferring 3D representations from general-purpose generative models to specialized medical tasks.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100259"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977037","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spatiotemporal dynamics of TMS-Evoked responses: A dual damped sine model analysis of cortical site and stimulation condition effects","authors":"Damián Jan","doi":"10.1016/j.neuri.2025.100254","DOIUrl":"10.1016/j.neuri.2025.100254","url":null,"abstract":"<div><h3>Background</h3><div>Transcranial magnetic stimulation combined with EEG (TMS-EEG) provides a non-invasive window into cortical excitability and connectivity. However, interpreting TMS-evoked potentials (TEPs) remains challenging due to pervasive artifacts and the limited physiological interpretability of descriptive analytical approaches.</div></div><div><h3>New method</h3><div>We introduce the Dual Damped Sine (DDS) model, a parametric framework that decomposes TEPs into physiologically meaningful parameters: amplitudes (A₁, A₂), frequencies (f₁, f₂), and damping constants (γ₁, γ₂). We applied DDS to the publicly available OpenNeuro dataset ds001849 to assess its ability to capture site- and condition-specific cortical responses.</div></div><div><h3>Results</h3><div>DDS achieved excellent model fits (median R² ≈ 0.95; RMSE ≤10⁻⁶) and revealed significant site- and condition-specific differences in the early TEP window (15–80 ms). Active TMS produced larger amplitudes and stronger damping, particularly at DLPFC, with frequencies constrained to physiological bands. These findings are consistent with previous evidence that early TEP components reflect site-specific cortical activation (Siebner et al., 2019; Freedberg et al., 2020).</div><div><strong>Comparison with existing methods</strong>:While traditional similarity metrics quantify global waveform differences, DDS provides mechanistic interpretation of TEP dynamics through its parametric decomposition. The model captures how cortical responses evolve in time, offering insights into excitatory-inhibitory dynamics.</div></div><div><h3>Conclusions</h3><div>DDS represents a novel analytical approach that not only confirms established findings about early TEP specificity but also provides physiologically interpretable parameters describing cortical response dynamics. This parametric framework advances TMS-EEG methodology by bridging the gap between waveform analysis and neurophysiological interpretation.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100254"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CleanEEG: A U-Net based deep learning framework for robust EEG artifact removal","authors":"Dhruva P. Achar ,&nbsp;Shavantrevva Bilakeri ,&nbsp;Karunakar A. Kotegar ,&nbsp;Kurupath Radhakrishnan","doi":"10.1016/j.neuri.2026.100264","DOIUrl":"10.1016/j.neuri.2026.100264","url":null,"abstract":"<div><div>High-frequency oscillations (HFOs) are vital biomarkers for identifying the seizure onset zone (SOZ) in patients with drug-resistant epilepsy (DRE). However, EEG artifacts especially muscle and power-line noise overlapping with the HFO frequency range (80–250 Hz) pose significant challenges for accurate detection. Traditional artifact removal methods like independent component analysis (ICA) are labor-intensive and subjective, highlighting the need for automated pre-processing techniques. This study introduces CleanEEG, a U-Net based encoder–decoder model designed to automate artifact removal from clinical EEG. CleanEEG was trained on paired noisy and clean sleep EEG segments from 25 DRE patients (177 segment pairs) at a 512 Hz sampling rate, with clean targets generated through ICA pre-processing. Model performance was quantitatively evaluated on an independent validation set comprising 24 segment pairs from six separate patients excluded from training. Evaluation metrics included relative root mean square error (RRMSE), correlation coefficient (CC), and signal-to-noise ratio (SNR). CleanEEG effectively removed muscle and power-line noise artifacts while preserving important clinical features such as interictal epileptiform discharges (IEDs) and brief potentially ictal rhythmic discharges (BIRDs). The model significantly improved signal quality across electrodes, reducing reconstruction errors and increasing SNR. Additionally, CleanEEG preserved neural activity without introducing distortions and qualitatively demonstrated artifact removal capability on unseen awake EEG data. In a representative DRE patient, critical spatial patterns of HFOs were maintained, essential for accurate SOZ localization. Overall, CleanEEG offers an automated, robust, and efficient solution for artifact removal, enhancing diagnostic accuracy in epilepsy monitoring and HFO analysis, particularly in long-term scalp EEG recordings.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100264"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146187320","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From epilepsy seizure classification to detection: A deep learning-based approach for raw EEG signals","authors":"Davy Darankoum ,&nbsp;Manon Villalba ,&nbsp;Clélia Allioux ,&nbsp;Baptiste Caraballo ,&nbsp;Carine Dumont ,&nbsp;Eloïse Gronlier ,&nbsp;Corinne Roucard ,&nbsp;Yann Roche ,&nbsp;Chloé Habermacher ,&nbsp;Sergei Grudinin ,&nbsp;Julien Volle","doi":"10.1016/j.neuri.2026.100263","DOIUrl":"10.1016/j.neuri.2026.100263","url":null,"abstract":"<div><div>Epilepsy is the most prevalent neurological disorder in the world. Although epilepsy has been recognized for centuries, clinical doctors still lack reliable automated tools to diagnose epileptic seizures in electroencephalograms (EEGs). The research community has made significant efforts to develop automated systems for identifying and quantifying epileptic seizures, with many studies reporting excellent accuracy. However, clinicians continue to rely on manual annotations because automated techniques exhibit poor generalization performance when applied to EEG data from new patients. Another challenge in the field is translating the results of preclinical studies conducted on animals to clinical applications in humans.</div><div>This work contributes to both challenges. Firstly, we investigate the reasons behind the lack of generalization in automatic models. We find that most existing techniques are evaluated on seizure classification tasks, while clinical doctors primarily encounter detection tasks in their practice. We demonstrate that the performance of automated pipelines differs significantly between the two and identify the key distinction between the tasks: classification presumes a prior separation between seizure and non-seizure EEG signals, whereas detection requires no such prior knowledge. Secondly, we bridge the gap between preclinical and clinical studies by developing novel deep learning architectures. Our best model, trained on EEG data from epileptic mice, demonstrates excellent generalization with an F1-score of 93% when tested on human data.</div></div>","PeriodicalId":74295,"journal":{"name":"Neuroscience informatics","volume":"6 1","pages":"Article 100263"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146187319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}