Journal of Neuroscience Methods最新文献

{"title":"Validation of structure tensor analysis for orientation estimation in brain tissue microscopy.","authors":"Bryson Gray, Andrew Smith, Allan MacKenzie-Graham, David W Shattuck, Daniel J Tward","doi":"10.1016/j.jneumeth.2025.110539","DOIUrl":"https://doi.org/10.1016/j.jneumeth.2025.110539","url":null,"abstract":"<p><strong>Background: </strong>Accurate localization of white matter pathways using diffusion MRI is critical to investigating brain connectivity, but the accuracy of current methods is not thoroughly understood. A fruitful approach to validating accuracy is to consider microscopy data that have been co-registered with MRI of post mortem samples. In this setting, structure tensor analysis is a standard approach to computing local orientations. However, structure tensor analysis itself has not been well-validated and is subject to uncertainty in its angular resolution, and selectivity to specific spatial scales.</p><p><strong>New method: </strong>Here, we conducted a simulation study to investigate the accuracy of using structure tensors to estimate the orientations of fibers arranged in configurations with and without crossings.</p><p><strong>Results: </strong>We examined a range of simulated conditions, with a focus on investigating the method's behavior on images with anisotropic resolution, which is particularly common in microscopy data acquisition. We also analyzed 2D and 3D optical microscopy data.</p><p><strong>Comparison with existing methods: </strong>Our results show that parameter choice in structure tensor analysis has relatively little effect on accuracy for estimating single orientations, although accuracy decreases with increasing resolution anisotropy. On the other hand, when estimating the orientations of crossing fibers, the choice of parameters becomes critical, and poor choices result in orientation estimates that are essentially random.</p><p><strong>Conclusions: </strong>This work provides a set of recommendations for researchers seeking to apply structure tensor analysis effectively in the study of axonal orientations in brain imaging data and quantifies the method's limitations, particularly in the case of anisotropic data.</p>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":" ","pages":"110539"},"PeriodicalIF":2.3,"publicationDate":"2025-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144753616","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effects of focused ultrasound stimulation on various in vitro neurological cell models","authors":"Iqra Bano ,&nbsp;Jaison Jeevanandam ,&nbsp;Grygoriy Tsenov","doi":"10.1016/j.jneumeth.2025.110544","DOIUrl":"10.1016/j.jneumeth.2025.110544","url":null,"abstract":"<div><div>Focused ultrasound stimulation (FUS) is rapidly gaining attention as a non-invasive and highly precise neuromodulatory technique with broad therapeutic potential in neurological and psychiatric disorders. While most reviews to date have emphasized in vivo and clinical studies, the cellular mechanisms underlying FUS remain underexplored. This study presents an innovative and thorough synthesis of FUS effects in in vitro neurological cell models, including SH-SY5Y, PC12, BV2 microglia, NSC-34 motor neurons, and human iPSC-derived neurons and astrocytes. These models offer essential insights into the mechanisms by which FUS influences intracellular calcium dynamics, mitigates oxidative stress, modulates inflammatory responses, and stimulates autophagy, thus facilitating neuroprotection and synaptic resilience in various disease contexts, including Parkinson’s disease, Alzheimer’s disease, schizophrenia, epilepsy, multiple sclerosis, OCD, and traumatic brain injury. Mapping disease-specific results with comprehensive FUS sonication parameters, this evaluation only focuses on cell-based systems, which is a fundamental advance. Additionally, it emphasizes the incorporation of new technology into FUS, such as acoustically responsive biomaterials, microbubble-assisted gene transfection, and nanoparticle-mediated medication delivery. The study highlights the increasing importance of AI in directing real-time FUS targeting and optimizing parameters, which is leading to tailored neuromodulation treatments. This study establishes a solid groundwork for the advancement of FUS in preclinical research by connecting the dots between cellular bioeffects and translational potential. It highlights the critical need for multidisciplinary methods, standardization, and the use of 3D organoid systems for next-generation brain treatments that fully use FUS.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110544"},"PeriodicalIF":2.3,"publicationDate":"2025-07-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144722687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Closed-loop phase-targeted stimulation during sleep: Open-source benchmarking of methods and a novel algorithm for the epileptic brain","authors":"Vicki Li ,&nbsp;Simeon M. Wong ,&nbsp;George M. Ibrahim","doi":"10.1016/j.jneumeth.2025.110543","DOIUrl":"10.1016/j.jneumeth.2025.110543","url":null,"abstract":"<div><h3>Background</h3><div>Phase-targeted auditory stimulation (PTAS) during sleep has been shown to enhance slow oscillations (SOs) and improve memory consolidation through closed-loop delivery of auditory stimuli at the up-phase of SOs. However, clinical translation of PTAS therapy has been hindered by challenges in the estimation of real-time phase. Our scoping review of 53 PTAS studies identified substantial variability in phase estimation methods and therapeutic outcomes. In particular, there were no validated methods for clinical populations with pathological electroencephalography (EEG) features, such as persons with epilepsy, where interictal epileptiform discharges (IEDs) compromise the performance of real-time PTAS delivery.</div></div><div><h3>New method</h3><div>To address critical limitations in the application of existing approaches to the epileptic brain, we developed TWave, a real-time algorithm that integrates wavelet-based phase estimation with predictive modelling and multi-feature validation. TWave is designed to maintain SO phase estimation performance while rejecting pathological EEG artifacts to achieve the temporal precision required for effective PTAS.</div></div><div><h3>Results</h3><div>TWave achieved high phase estimation accuracy and precision in healthy adult (mean error=0.11 radians; SD=1.23 radians) and paediatric epilepsy (mean error=0.26 radians; SD=1.22 radians) EEG recordings. Importantly, TWave successfully rejected 83 % of IEDs while maintaining sensitivity to SOs.</div></div><div><h3>Comparison with existing algorithms</h3><div>Benchmarking against four commonly used algorithms demonstrated TWave’s superior performance in maintaining phase estimation precision across normative and epilepsy EEG recordings.</div></div><div><h3>Conclusion</h3><div>The current work accelerates clinical translation of PTAS by providing a validated approach to real-time phase estimation and providing an open-source toolbox to increase reproducibility in sleep modulation research.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110543"},"PeriodicalIF":2.3,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144717967","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reproducibility of HERMES-measured GABA+ and glutathione in the mesial temporal lobe","authors":"Marilena M. DeMayo ,&nbsp;Mary Botros ,&nbsp;Tiffany K. Bell ,&nbsp;Mark Mikkelsen ,&nbsp;Victoria Mosher ,&nbsp;Antis George ,&nbsp;Alexander McGirr ,&nbsp;Paolo Federico ,&nbsp;Ashley D. Harris","doi":"10.1016/j.jneumeth.2025.110542","DOIUrl":"10.1016/j.jneumeth.2025.110542","url":null,"abstract":"<div><h3>Background</h3><div>There is growing interest in using Hadamard Encoding and Reconstruction for MEGA-Edited Spectroscopy (HERMES) within the mesial temporal lobe (MTL). For cross-sectional group comparisons and longitudinal repeated measures designs, an understanding of the internal and test-retest validity of γ-aminobutyric acid (GABA+) and glutathione (GSH) is critical. We therefore evaluated the reproducibility of the consensus recommended semi-localization by adiabatic selective refocusing (sLASER) localization for edited-MRS acquisitions in a challenging region, the MTL.</div></div><div><h3>New method</h3><div>Data were acquired in 15 participants. Single voxel HERMES was collected in the left MTL (two acquisitions) and the right MTL (one acquisition). Participants were repositioned between the two left HERMES acquisitions. An ANOVA was used to assess differences between acquisitions. To assess measurement variation in the repeated left of GABA+ and GSH measures within the left MTL difference values and coefficients of variation (CVs) were calculated.</div></div><div><h3>Results</h3><div>There were no significant differences in metabolite values between any of the acquisitions. The mean difference between the metabolite measures from the repeated left acquisitions centred close to zero, and the average CVs were 14.09 % for GABA+ and 18.94 % for GSH.</div></div><div><h3>Comparison with existing methods</h3><div>The CVs of GABA+ and GSH in the MTL obtained from a HERMES acquisition were comparable to GABA+ or GSH-edited acquisitions in this region, and to data from cortical voxels using HERMES acquisitions.</div></div><div><h3>Conclusions</h3><div>This supports the use of HERMES in the MTL, a challenging region for MRS. However, larger samples and caution in interpretation may be required in repeated-measures designs.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110542"},"PeriodicalIF":2.7,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144713150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimized methyl green-pyronin Y staining for layer visualization in frozen mouse cerebellum","authors":"Hyeju Kim,&nbsp;Soo-Jong Um","doi":"10.1016/j.jneumeth.2025.110540","DOIUrl":"10.1016/j.jneumeth.2025.110540","url":null,"abstract":"<div><h3>Background</h3><div>Conventional histological stains, such as hematoxylin and eosin (H&amp;E) or toluidine blue O (TBO), have a limited ability to clearly delineate the layered architecture of the cerebellar cortex.</div></div><div><h3>New method</h3><div>We applied methyl green–pyronin Y (MGP) staining, which is traditionally used for nucleic acid differentiation, to frozen mouse cerebellar sections to enhance visualization of cortical layers and neuronal subtypes.</div></div><div><h3>Results</h3><div>MGP staining yielded strong contrast between cell types: Purkinje cells stained distinctly pink, while granule cells appeared green. This enabled clear identification of cerebellar lamination and neuronal distribution.</div></div><div><h3>Comparison with existing methods</h3><div>In H&amp;E or TBO staining, Purkinje and granule cells are colored similarly, which obscures layer boundaries. Although immunohistochemistry is commonly used to distinguish these cell types, MGP staining provides a rapid, color-based distinction without the need for antibodies or fluorescence.</div></div><div><h3>Conclusions</h3><div>MGP staining provides a fast and cost-effective alternative for analyzing cerebellar tissue, enabling clear visualization of cortical layering and facilitating the morphological screening of cerebellar abnormalities.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110540"},"PeriodicalIF":2.7,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144703372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A new gradient-based method for analyzing brain white matter fiber geometry","authors":"Jiaolong Qin ,&nbsp;Weihong Dong ,&nbsp;Huangjing Ni ,&nbsp;Ye Wu ,&nbsp;Haiyan Liu ,&nbsp;Zhijian Yao ,&nbsp;Qing Lu","doi":"10.1016/j.jneumeth.2025.110538","DOIUrl":"10.1016/j.jneumeth.2025.110538","url":null,"abstract":"<div><h3>Background</h3><div>Precise geometric and morphometric analyses of brain fiber pathways are crucial for unraveling brain organization and mechanisms underlying normal and pathological brain functions. However, existing methods for white matter (WM) fiber geometry analysis remain limited.</div></div><div><h3>New method</h3><div>We propose a novel Large-scale Gradient Feature (LsGF) metric to quantify the tangent direction change rate along fiber streamlines. Using intra-class correlation coefficients (ICC), we systematically evaluated the stability of LsGF maps under two key factors: streamline count and tractography algorithm. LsGF was then applied to investigate gender disparities in WM morphology, with sensitivity assessed by comparing LsGF maps against fiber length maps.</div></div><div><h3>Results</h3><div>Results showed that LsGF exhibited remarkable robustness to variations in streamline count (99 % of ICCs &gt; 0.8), but demonstrated significant dependency on tractography algorithm (less than 60 % of ICCs &gt; 0.6). Application of the LsGF method to gender dimorphism analysis uncovered distinct geometric patterns primarily in the thalamus, internal capsule, cerebellum, corpus callosum, lingual gyrus, fusiform gyrus, precuneus, gyrus rectus, orbitofrontal cortex, cingulate cortex, calcarine, and olfactory regions.</div></div><div><h3>Comparison with existing methods</h3><div>Comparative analysis indicated that LsGF outperformed fiber length metrics in detecting microstructural geometric complexity, whereas the latter more effectively characterized macroscale architecture features. These findings underscore the complementary value of LsGF and fiber length metric in WM analysis.</div></div><div><h3>Conclusions</h3><div>The LsGF map enables voxel-wise analysis of quantitative streamline metrics across the whole brain, highlighting the necessity of consistent tractography methods for reliable results.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110538"},"PeriodicalIF":2.7,"publicationDate":"2025-07-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144675003","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hybrid BCI for upper limb rehabilitation: integrating MI with peripheral field SSVEP stimulation","authors":"Ruoqing Zhang ,&nbsp;Zhaohui Li ,&nbsp;Xiaohu Pan ,&nbsp;Hongyan Cui ,&nbsp;Xiaogang Chen","doi":"10.1016/j.jneumeth.2025.110537","DOIUrl":"10.1016/j.jneumeth.2025.110537","url":null,"abstract":"<div><h3>Background</h3><div>Rehabilitation systems based on brain-computer interfaces (BCIs) hold significant potential for stroke patients. Existing systems, predominantly relying on motor imagery (MI), have room for improvement in both performance and user comfort. This study aims to enhance these aspects by developing a hybrid BCI system integrating MI with steady-state visual evoked potentials (SSVEPs) elicited by peripheral visual field stimulation.</div></div><div><h3>New methods</h3><div>The system is coupled with a soft robotic hand for feedback, forming a closed-loop framework. The design incorporates concentric rings with 7° and 10° eccentricities as peripheral stimuli, flashing at frequencies of 34 Hz and 35 Hz for left and right sides, respectively, to evoke SSVEPs. A central video (304 ×304 pixels) of left-hand/right-hand grasping motions guides subjects in performing synchronized MI tasks simply by focusing on it, which could also complete the SSVEP task.</div></div><div><h3>Results</h3><div>The offline results of 11 subjects showed that the classification result of MI was 70.65 ± 3.38 %, and the SSVEP result was 96.04 ± 3.33 %, and the fusion result reached 96.23 ± 3.21 %, which confirmed the validity of the fusion method. The online experiment of 11 subjects achieved a result of 97.12 ± 2.09 %, validating the feasibility of the system.</div></div><div><h3>Comparison with existing methods</h3><div>The proposed system improves the comfort level while ensuring the performance of the system as compared to the existing systems.</div></div><div><h3>Conclusion</h3><div>The feasibility of the proposed system was verified by offline and online experiments to advance the clinical applications.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110537"},"PeriodicalIF":2.7,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144655297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-class mental Task Classification based Brain-Computer Interface using Improved Remora depthwise convolutional adaptive neuro-fuzzy inference network model","authors":"D. Deepika ,&nbsp;G. Rekha","doi":"10.1016/j.jneumeth.2025.110536","DOIUrl":"10.1016/j.jneumeth.2025.110536","url":null,"abstract":"<div><h3>Background</h3><div>Brain-computer interfaces (BCIs) offer a promising avenue for individuals with severe motor disabilities to interact with the world. By decoding brain signals, BCIs can enable users to control devices and communicate thoughts. However, challenges such as noise in EEG signals and limited data availability hinder the development of accurate and reliable BCI systems. Nonetheless, problems persist, including limited data availability, noisy EEG signals, real-time performance limitations, and reduced classification accuracy.</div></div><div><h3>New method</h3><div>To overcome this, the present work proposes an efficient Multi-Class Mental Task Classification based BCI using deep learning techniques. Initially, the obtained EEG data is pre-processed with a Finite Linear Haar wavelet-based Filtering (FLHF) technique to remove disturbances in EEG data. Afterwards, optimal feature extraction utilizes a Hybrid dynamic centre binary pattern and multi-threshold-based ternary pattern (H-DCBP-MTTP) technique to extract characteristics from pre-processed EEG data. Finally, the Improved Remora depthwise convolutional adaptive neuro-fuzzy inference network (IRDCANFIN) model is used to classify the mental tasks. To improve classification results, the model’s parameters are fine-tuned using an Improved Remora optimization approach (IROA).</div></div><div><h3>Results</h3><div>The proposed approach’s performance is examined using the BCI laboratory dataset and the EEG Psychiatric Disorders Dataset, which yield accuracy results of 99.3 % and 99.56 %, respectively. Furthermore, evaluation results show that the proposed approach outperforms existing models.</div></div><div><h3>Comparison with existing methods</h3><div>Compared to existing models, such as DQN with a 1D-CNN-LSTM, GSP-ML, Shallow 1D-CNN, KNN, and SVM, and the proposed approach yields effective results in terms of accuracy, robustness, and computational efficiency.</div></div><div><h3>Conclusion</h3><div>The proposed IRDCANFIN classifier is used to classify multiple classes of mental tasks like baseline, counting, multiplication, letter composing, and rotation.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110536"},"PeriodicalIF":2.3,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144667771","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep clustering of polysomnography data to characterize sleep structure in healthy sleep and non-rapid eye movement parasomnias","authors":"I.A.M. Huijben ,&nbsp;R.J.G. van Sloun ,&nbsp;A. Pijpers ,&nbsp;S. Overeem ,&nbsp;M.M. van Gilst","doi":"10.1016/j.jneumeth.2025.110516","DOIUrl":"10.1016/j.jneumeth.2025.110516","url":null,"abstract":"<div><h3>Background:</h3><div>The clinical standard to interpret polysomnography (PSG) data is to categorize sleep in five stages, which omits information. SOM-CPC is an unsupervised method that extracts features through contrastive predictive coding (CPC), and visualizes them in two dimensions using a self-organizing map (SOM). We propose various visualizations and analyses for pattern recognition in PSG data through SOM-CPC.</div></div><div><h3>New method:</h3><div>We used SOM-CPC to learn a representation of 30-s multi-channel epochs from two datasets of healthy sleepers (<span><math><mrow><mi>n</mi><mo>=</mo><mn>52</mn></mrow></math></span> and <span><math><mrow><mi>n</mi><mo>=</mo><mn>22</mn></mrow></math></span> in the test sets). SOM-CPC was, additionally, used to further characterize awakenings from slow wave sleep (SWS) in non-rapid eye movement (NREM) parasomnias. For the latter, SOM-CPC was trained on 5-s single-channel EEG windows of non-rapid eye movement parasomnias and matched healthy controls (test set: <span><math><mrow><mi>n</mi><mo>=</mo><mn>67</mn></mrow></math></span>).</div></div><div><h3>Results:</h3><div>SOM-CPC organized epochs of healthy sleepers such that it separated sleep stages, and also encoded age of the subjects and time in the night. Parasomnia episodes, compared to non-behavioral SWS awakenings, exhibited higher SWS-specificity prior to transition to wakefulness, higher Wake-specificity post-transition, and longer durations.</div></div><div><h3>Comparison with existing methods:</h3><div>The learned representations were compared against gold-standard sleep stage labels and variables known to impact sleep structure.</div></div><div><h3>Conclusions:</h3><div>SOM-CPC seems a useful model for pattern discovery in PSG data, as it enables observation of state changes that are more intricate than full sleep stage transitions. It, moreover, provided further evidence for signal level differences in the EEG between SWS awakenings with and without parasomnia episodes.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110516"},"PeriodicalIF":2.7,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144649736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CSCE: Cross Supervising and Confidence Enhancement pseudo-labels for semi-supervised subcortical brain structure segmentation","authors":"Yuan Sui,&nbsp;Yujie Zhang,&nbsp;Chengan Liu","doi":"10.1016/j.jneumeth.2025.110522","DOIUrl":"10.1016/j.jneumeth.2025.110522","url":null,"abstract":"<div><div>Robust and accurate segmentation of subcortical structures in brain MR images lays the foundation for observation, analysis and treatment planning of various brain diseases. Deep learning techniques based on Deep Neural Networks (DNNs) have achieved remarkable results in medical image segmentation by using abundant labeled data. However, due to the time-consuming and expensive of acquiring high quality annotations of brain subcortical structures, semi-supervised algorithms become practical in application. In this paper, we propose a novel framework for semi-supervised subcortical brain structure segmentation, based on pseudo-labels <strong>C</strong>ross <strong>S</strong>upervising and <strong>C</strong>onfidence <strong>E</strong>nhancement (CSCE). Our framework comprises dual student-teacher models, specifically a U-Net and a TransUNet. For unlabeled data training, the TransUNet teacher generates pseudo-labels to supervise the U-Net student, while the U-Net teacher generates pseudo-labels to supervise the TransUNet student. This mutual supervision between the two models promotes and enhances their performance synergistically. We have designed two mechanisms to enhance the confidence of pseudo-labels to improve the reliability of cross-supervision: a) Using information entropy to describe uncertainty quantitatively; b) Design an auxiliary detection task to perform uncertainty detection on the pseudo-labels output by the teacher model, and then screened out reliable pseudo-labels for cross-supervision. Finally, we construct an end-to-end deep brain structure segmentation network only using one teacher network (U-Net or TransUNet) for inference, the segmentation results are significantly improved without increasing the parameters amount and segmentation time compared with supervised U-Net or TransUNet based segmentation algorithms. Comprehensive experiments are performed on two public benchmark brain MRI datasets. The proposed method achieves the best Dice scores and MHD values on both datasets compared to several recent state-of-the-art semi-supervised segmentation methods.</div></div>","PeriodicalId":16415,"journal":{"name":"Journal of Neuroscience Methods","volume":"423 ","pages":"Article 110522"},"PeriodicalIF":2.7,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144623475","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}