Medical & Biological Engineering & Computing最新文献

{"title":"Unsupervised functional stage detection from EEG using topological data analysis.","authors":"Aleksandr Abramov, Ekaterina Mikhaylets, Vsevolod Chernyshev, Alexander Kaplan, Fedor Ratnikov","doi":"10.1007/s11517-026-03683-0","DOIUrl":"https://doi.org/10.1007/s11517-026-03683-0","url":null,"abstract":"<p><p>Electroencephalogram (EEG) segmentation is still largely performed manually by specialists. Recent studies have proposed the state-detecting algorithm (SDA) - an automated technique for detecting functional stages with minimal human intervention - which has proven effective using spectral features. This paper introduces an alternative feature engineering approach that explores the spatial structure of data using the emerging field of topological data analysis (TDA). We propose an algorithm to extract a comprehensive topological description of EEG with tens of thousands of features and introduce the quick state-detecting algorithm (QSDA), a feature selection framework based on SDA performance. Next, we apply the procedure to EEG recordings from two distinct domains - Guhyasamaja Tantra meditation and whole-night polysomnographic sleep - and observe well-separated stage boundaries across most recordings, suggesting that TDA may complement or outperform spectral features in some settings. We also observed encouraging performance on noisy data, whereas shuffled epochs produced substantially weaker boundaries, supporting the robustness of the technique within the settings examined. Overall, the algorithm appears to be a promising feature engineering approach for SDA and suggests that the spatial structure of EEG may contain informative patterns. These findings motivate further investigation of their possible physiological interpretation, as well as an assessment across broader EEG paradigms.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149129004","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 learning-based automated scanline localization in suprahyoid muscle ultrasound for quantitative dysphagia assessment.","authors":"Seyeon Kim, Junbeom Lee, Sinwoo Kim, Seung-Ho Shin, Jaewoong Kang, Joo Hye Sung, Chulho Kim, Jae Jun Lee, Taikyeong Jeong","doi":"10.1007/s11517-026-03629-6","DOIUrl":"https://doi.org/10.1007/s11517-026-03629-6","url":null,"abstract":"<p><p>Ultrasound-based assessment of suprahyoid muscle (SHM) motion has emerged as a promising approach for evaluating dysphagia. However, quantitative analysis remains limited by operator-dependent scanline placement and variability in manual interpretation. This study proposes an integrated framework combining mathematical modeling, automated SHM segmentation, scanline determination, and deep learning-based analysis for objective dysphagia assessment. A total of 371 ultrasound images acquired from patients with normal, mild, and severe dysphagia were analyzed. Clinically relevant SHM-derived variables, including thickness, displacement, SHM Difference (Δ = D - T), and total duration, were incorporated into a mathematical interpretation framework. U-Net was employed for SHM segmentation and automatic scanline determination, and agreement with expert-selected scanlines was evaluated using median-based accuracy, intraclass correlation coefficient (ICC), mean pairwise absolute difference (MPAD), and population standard deviation (SDpop). In addition, five deep learning architectures (ResNet-101, Fast R-CNN, YOLO11, VGG-19, and U-Net) were compared for dysphagia severity classification. The proposed framework demonstrated high agreement with expert assessments. U-Net achieved an average intersection-over-union (IoU) of 0.9599 across the three major SHM components, indicating near-expert segmentation performance. Automated scanline determination achieved an overall agreement of 72.3% with expert consensus, while manual scanline placement exhibited relatively low interobserver reliability (ICC = 0.386). Among the evaluated classification models, Fast R-CNN achieved the highest performance with an F1-score of 0.993 and an inference time of 3.44 ms. Grad-CAM analysis further confirmed that model attention was concentrated on clinically relevant SHM regions. The proposed framework provides an objective and reproducible approach for ultrasound-based dysphagia assessment by integrating mathematical modeling with AI-driven segmentation and classification. Automated SHM localization and scanline determination reduce operator dependency while maintaining expert-level accuracy. These findings support the potential of AI-assisted ultrasound as a clinically applicable tool for standardized dysphagia evaluation and real-time decision support.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149129023","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":"Reduced-order analysis for gait kinematics of paraplegic patients based on spatiotemporal mode decomposition.","authors":"Xin Wang, Yunzhe Zhang, Xi Zhang, Boyang Zhang, Jianqiao Guo","doi":"10.1007/s11517-026-03688-9","DOIUrl":"https://doi.org/10.1007/s11517-026-03688-9","url":null,"abstract":"<p><p>Complex human motion can be regarded as a summation of simple motion patterns. Existing studies have described kinematic synergy patterns with principal component analysis (PCA), but it overlooks time-frequency characteristics of gait. This study establishes a reduced-order spatiotemporal framework to obtain nonlinear kinematic patterns. Three paraplegic patients and twelve healthy subjects are recruited to conduct gait experiments. In the spatial domain, kinematic synergy modes of lower-limb joints are extracted by PCA. The principal modes at the instants of paraplegic patients' prominent joint torques are determined by calculating the contribution rate of weight. In the temporal domain, weights of principal modes are decomposed into nonlinear intrinsic mode functions (IMFs) using empirical mode decomposition (EMD). The obtained kinematic synergy modes of healthy subjects and the mild patient are consistent with results obtained from dynamic mode decomposition (DMD). An additional principal mode is observed in one of the analyzed paraplegic participants and dominated at the instants of prominent joint torque. The dominant frequency of the principal IMF for this participant is 5.44 times the stride frequency. The proposed PCA-EMD framework provides a quantitative description of patient-specific kinematic synergy patterns and altered temporal movement dynamics in gait.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117572","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":"Segmenting scaffolds with ingrown bone in micro-CT images: considerations of evaluation metrics and loss functions.","authors":"Sohaila Aboutaleb, Nellie Haug, Prachi Keni-McCray, Arthur R C McCray, Heidi Phillips, Stephanie Sharping, David Cohen, Julián Norato, Amy Wagoner Johnson","doi":"10.1007/s11517-026-03686-x","DOIUrl":"https://doi.org/10.1007/s11517-026-03686-x","url":null,"abstract":"<p><p>U-Net models are useful for biomedical image segmentation because they do not require large training sets. Here, we trained three U-Net models, each with a different loss function, to segment micro-CT image patches of scaffolds with ingrown bone to quantify bone growth fraction. Segmentation performance is typically reported using metrics like Accuracy, DSC, Precision, and Recall. However, these metrics did not reflect performance relative to the application's key output measure. We therefore aimed to establish an application-specific parameter, ΔBGF: the difference in bone growth fraction (BGF) between predicted output (PO) and ground truth (GT). Bland-Altman analysis showed small systematic bias (< 1.5 percentage points), supporting PO-GT agreement. Accuracy was the metric most strongly correlated to |ΔBGF| in nearly all comparisons (Steiger test), and this relationship was stable under bootstrap resampling. The models were robust to changes in orientation, resolution, and contrast, supporting broader generalizability. The cross-entropy loss model generated the most reliable segmentation. PO-GT error was small relative to meaningful biological differences, suggesting it is unlikely to affect future treatment comparisons. Overall, standard metrics can be misleading, while ΔBGF provides a reliable, application-specific measure of performance for this still developing use of U-Net in quantifying bone regeneration.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117647","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 two-stage PET/MRI attenuation-correction workflow for pseudo-CT and AC PET synthesis with GUI-based inference.","authors":"Seyede Sajede Mousavi Eshkiki, Alireza Sadremomtaz","doi":"10.1007/s11517-026-03685-y","DOIUrl":"https://doi.org/10.1007/s11517-026-03685-y","url":null,"abstract":"<p><p>Quantitative positron emission tomography/magnetic resonance imaging (PET/MRI) requires reliable attenuation correction, but MRI does not directly encode attenuation-related tissue information. This study implemented and internally evaluated a graphical user interface (GUI)-supported two-stage brain PET/MRI workflow for pseudo-computed tomography (pseudo-CT) generation and PET attenuation-correction synthesis using the CERMEP-IDB-MRXFDG cohort. The objective was not to introduce a new neural-network architecture, but to evaluate an attenuation-informed workflow that sequentially combines MRI-based pseudo-CT generation, attenuation-map estimation, and NAC PET correction. Thirty-seven normal adult subjects were analyzed using subject-level five-fold cross-validation. A first-stage two-dimensional U-Net predicted pseudo-CT from T1-weighted MRI. The pseudo-CT output was converted into an approximate 511-keV linear attenuation coefficient map (μ-map) and combined with non-attenuation-corrected (NAC) PET as input to a second U-Net for predicted attenuation-corrected (AC) PET synthesis. Patient-level pseudo-CT errors, image-similarity metrics, Bland-Altman agreement, and exploratory standardized uptake value (SUV) metrics were evaluated. The pseudo-CT model achieved whole-image mean absolute error (MAE) of 72.26 ± 20.11 HU and structural similarity index measure (SSIM) of 0.881 ± 0.024, with bone-dominated residual error. Predicted AC PET achieved 76.02% lower normalized MAE and 73.33% lower normalized root mean square error (RMSE) than NAC PET relative to reference AC PET. Exploratory SUV analysis showed mean percentage errors of 8.66% for SUVmean and 22.08% for SUVmax, with greater inter-subject variability in SUVmax. These findings support internal CERMEP feasibility, while nested and external validation remain necessary.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117438","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":"Path-dependent viscoelastic-hyperelastic response of brain tissue under multistage shear loading.","authors":"Hadi Nosrati, Mehdi Shafieian, Kurosh Darvish","doi":"10.1007/s11517-026-03692-z","DOIUrl":"https://doi.org/10.1007/s11517-026-03692-z","url":null,"abstract":"<p><p>Advancing computational models of repetitive traumatic brain injury (rTBI) requires understanding how prior loading alters the mechanical response of brain tissue. However, most previous studies have relied on single-stage loading or limited preconditioning protocols that do not capture path-dependent behavior. To bridge this gap, multistage simple shear experiments were performed on 14 bovine brain specimens. Successive ramp-and-hold and cyclic loading protocols examined the effects of loading mode, strain amplitude, and loading sequence within the reported sub-damaging loading range. Mechanical behavior was quantified using the shear modulus (μ) and Prony-series viscoelastic parameters (g₁-g₃, g_∞). In the present experiments, prior cyclic loading generally reduced peak stress and shear modulus. From the initial to the final loading stage, μ decreased from 2.94 ± 0.94 to 2.49 ± 0.63 kPa, g₂ increased from 0.14 ± 0.19 to 0.19 ± 0.17, and g₃ decreased from 0.21 ± 0.08 to 0.16 ± 0.06 (all p < 0.05). In contrast, g₁ and g_∞ showed no significant changes (p > 0.05). Cumulative strain was not significantly correlated with final μ (Spearman ρ = - 0.31, p > 0.05), whereas protocol sequence was associated with differences in μ within the investigated multistage loading histories. These findings demonstrate path- and sequence-dependent brain tissue mechanics that may inform constitutive models for repeated loading scenarios, including those relevant to rTBI.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117555","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":"Robust cross-anatomy image registration leveraging keypoint priors and contour-based assessment.","authors":"Yuanhao Xu, Heng Zeng, Chaoran Xue, Hui Xu, Yuanyuan Chen","doi":"10.1007/s11517-026-03681-2","DOIUrl":"https://doi.org/10.1007/s11517-026-03681-2","url":null,"abstract":"<p><p>Accurate registration between lateral cephalometric radiographs and facial profile photographs is a prerequisite for integrated craniofacial analysis, yet it remains challenging due to inherent modality gaps and the absence of clinically validated evaluation criteria. This study presents a clinically-driven engineering framework that systematically integrates high-resolution keypoint localization with a weighted Procrustes-based two-dimensional similarity registration. The uniform scale factor accommodates global magnification differences between imaging modalities, while shearing and local non-rigid deformation are excluded. To ensure clinical interpretability, we incorporate expert-defined anatomical priors as weights into the transformation process and introduce an adaptive seed-driven contour tracking (ASDCT) algorithm for precise performance evaluation. Experiments on 198 paired clinical samples demonstrate that our framework achieves stable performance (Contour Distance of 3.52 pixels; Contour Dilation Overlap of 87%) and provides metrics that correlate strongly with expert clinical assessments. By unifying landmark-guided alignment and boundary-consistency evaluation, this framework offers a robust, reproducible, and anatomically reliable solution for cross-modal craniofacial integration, effectively bridging the gap between computational precision and clinical practice.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117605","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":"AMGFM: A multimodal fusion model for depression identification based on adversarial metric learning.","authors":"Jing Zhu, Aohan Zhang, Xiaowei Li, Yizhou Li, Bin Hu","doi":"10.1007/s11517-026-03676-z","DOIUrl":"https://doi.org/10.1007/s11517-026-03676-z","url":null,"abstract":"<p><p>Early detection of depression is essential for timely intervention, yet objective assessment remains difficult. Recent studies have explored multimodal physiological signals, including electroencephalography (EEG), eye movement, and pupil dynamics, to improve depression recognition. However, existing multimodal approaches still face two major limitations: insufficient alignment across heterogeneous modalities, which results in modality-dependent feature representations, and limited systematic analysis of how emotional stimuli and cross-modal interactions affect recognition performance, thereby constraining both accuracy and interpretability. To address these issues, we propose AMGFM, an Adversarial-Metric Graph Fusion Model for multimodal depression recognition. AMGFM uses adversarial learning to reduce global distribution discrepancies across modalities and metric learning to enforce intra-class compactness and inter-class separability in a shared latent space. In addition, a hierarchical graph fusion module explicitly models and adaptively integrates unimodal, bimodal, and trimodal interactions to exploit complementary information. Experiments on EFVP and AADP demonstrate that AMGFM consistently outperforms unimodal baselines and representative multimodal fusion approaches, achieving up to 90.33% accuracy. Further analyses indicate that neutral stimuli provide stronger discriminative cues than emotional stimuli, and that EEG serves as an effective target modality for aligning other signals.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117441","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":"HSF-Net: a hybrid-scale fusion network for intracranial aneurysm segmentation in CTA images.","authors":"Shuwen Yang, Peipei Wang, Mingquan Ye, Lele Liu, Yajie Li, Yunfeng Zhou","doi":"10.1007/s11517-026-03684-z","DOIUrl":"https://doi.org/10.1007/s11517-026-03684-z","url":null,"abstract":"<p><p>Rupture of intracranial aneurysms is a major cause of subarachnoid hemorrhage, making accurate segmentation important for aneurysm assessment and treatment planning. However, automated segmentation in computed tomography angiography (CTA) remains challenging because of small lesion size, complex morphology, and ambiguous boundaries between aneurysms and adjacent vessels. To address these challenges, we propose HSF-Net, a Swin UNETR-based segmentation network that integrates a Hybrid-Scale Fusion (HSF) module and the Convolutional Block Attention Module (CBAM) into the decoder. Unlike conventional decoders that mainly rely on simple feature concatenation, HSF-Net follows a \"fuse first, refine later\" strategy, in which HSF performs input-dependent multi-scale feature fusion and CBAM subsequently refines the fused features through channel and spatial attention. On the internal independent test set, HSF-Net achieved a mean patient-level Dice score of 83.4%, outperforming the Swin UNETR baseline and obtaining the highest Dice score among the evaluated methods under the same evaluation protocol. Ablation studies confirmed the complementary contributions of HSF and CBAM. Evaluation on an independent external cohort provided preliminary evidence of its cross-center applicability. These results suggest that HSF-Net provides an effective approach for automated intracranial aneurysm segmentation in CTA images.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117611","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":"LabCora: label correlation-aware semi-supervised multi-label medical image classification.","authors":"Yi Zhong, Zhiqiang Shen, Peng Cao, Chao Wan, Jinzhu Yang, Osmar R Zaiane","doi":"10.1007/s11517-026-03679-w","DOIUrl":"https://doi.org/10.1007/s11517-026-03679-w","url":null,"abstract":"<p><p>Semi-supervised learning has shown great potential in reducing the annotation burden for multi-label medical image classification. However, most existing methods suffer from confirmation bias due to unreliable pseudo-labels, while the distribution mismatch between labeled and unlabeled data frequently limits the model's generalization capability. In this paper, we propose LabCora, a novel framework for semi-supervised multi-label medical image classification. Specifically, we propose a Label Correlation-guided Pseudo-Labeling module, which mitigates confirmation bias by explicitly modeling label correlations and using them to correct unreliable pseudo-labels. In addition, a Cross-Domain Adversarial module is introduced to alleviate the distribution mismatch via feature-level adversarial training. By integrating these modules within a Co-training paradigm, label correlation-guided pseudo-label refinement and cross-domain adversarial alignment work together to enhance both pseudo-label reliability and feature generalization. Extensive experimental results on Chest X-Ray14, CheXpert, and ODIR-5K demonstrate that LabCora consistently outperforms state-of-the-art methods, particularly in low-label regimes.</p>","PeriodicalId":49840,"journal":{"name":"Medical & Biological Engineering & Computing","volume":" ","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117547","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}