Computer methods and programs in biomedicine最新文献

{"title":"Digital pathology-based prognostic model for hepatocellular carcinoma: Integrating pathomics signatures with clinical parameters for recurrence prediction and biological interpretation","authors":"Qi Wang ,&nbsp;Yuxi Huang ,&nbsp;Yu Zhang ,&nbsp;Yu Zhu ,&nbsp;Peng Hu ,&nbsp;Yongfu Xu ,&nbsp;Zhen-yu Jiang ,&nbsp;Long Liu ,&nbsp;Shao-wei Li","doi":"10.1016/j.cmpb.2025.109180","DOIUrl":"10.1016/j.cmpb.2025.109180","url":null,"abstract":"<div><h3>Background</h3><div>Hepatocellular carcinoma (HCC) remains a therapeutic challenge due to high post-resection recurrence rates and heterogeneous outcomes. We developed and validated a digital pathology-based prognostic model combining pathomics signatures with clinical parameters to predict recurrence and elucidate biological mechanisms.</div></div><div><h3>Methods</h3><div>In this multicenter retrospective study, 294 HCC patients (training set: <em>n</em> = 198; validation set: <em>n</em> = 96) undergoing curative hepatectomy were analyzed. Pathomics features were quantitatively extracted from H&amp;E-stained whole-slide images. Predictive modeling incorporated machine learning approaches (DT, KNN, LASSO, NB, RF, SVM) with clinical variables. Model performance was evaluated through ROC analysis, calibration, and decision curve analysis. Biological interpretation leveraged TCGA transcriptomic data analyzed via GSEA and WGCNA.</div></div><div><h3>Results</h3><div>Tumor and peri‑tumor pathomics parameters showed some complementarity in the prediction of HCC recurrence. The combined LASSO-based model showed the best predictive efficacy, with AUCs of 0.850 and 0.807 in the training and validation sets, respectively. The integrated pathomics-clinical model achieved AUCs of 0.893 and 0.860 in training and validation sets. Bioinformatics analysis suggested that the pathomics was correlated with the tumor immune microenvironment, as verified by multiple immunofluorescence staining of the validation set.</div></div><div><h3>Conclusion</h3><div>This study establishes a robust digital pathology framework that not only improves HCC recurrence prediction beyond conventional biomarkers but also provides mechanistic insights into tumor-immune crosstalk.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109180"},"PeriodicalIF":4.8,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145676357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ElasticMorph: Plug-and-play second-order elastic regularization for medical image registration","authors":"Zhaoxi Lin,&nbsp;Shan Jiang,&nbsp;Zeyang Zhou,&nbsp;Zhiyong Yang,&nbsp;Yuhua Li","doi":"10.1016/j.cmpb.2025.109182","DOIUrl":"10.1016/j.cmpb.2025.109182","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Deformable image registration is crucial for many clinical applications, but current learning-based methods often rely on simple smoothness losses. This can produce physically implausible deformations, such as tissue folding, and creates a trade-off between registration accuracy and anatomical correctness. The objective of this study is to develop and validate a novel, physics-informed regularizer that simultaneously improves both the accuracy and physical plausibility of learning-based registration with negligible computational overhead.</div></div><div><h3>Methods:</h3><div>We propose ElasticMorph, a plug-and-play loss function derived from the Navier–Cauchy equation. This regularizer penalizes both curvature and divergence in the deformation field. We derive an approximation based, measure theoretic bound under small strain assumptions, which links the elastic residual to an upper bound on the number of negative Jacobian voxels. This gives a principled and differentiable surrogate for folding control. The method was evaluated on two public brain MRI benchmarks (IXI and LPBA-40) by integrating it into four state-of-the-art CNN and transformer backbones. Performance was quantified using the Dice Similarity Coefficient (DSC) and the percentage of negative Jacobians (<span><math><mrow><mtext>%</mtext><mi>n</mi><mi>e</mi><mi>g</mi><mi>J</mi></mrow></math></span>), with statistical significance assessed via paired t-tests.</div></div><div><h3>Results:</h3><div>Across all tested backbones and datasets, ElasticMorph consistently improved registration accuracy, increasing the mean DSC by 0.56–4.92 %. Concurrently, it suppressed folding artifacts, reducing the <span><math><mrow><mtext>%</mtext><mi>n</mi><mi>e</mi><mi>g</mi><mi>J</mi></mrow></math></span> by 24%–42%. This was achieved with a minimal increase in training time <span><math><mrow><mo>≤</mo><mn>8</mn><mtext>%</mtext></mrow></math></span> and memory <span><math><mrow><mo>≤</mo><mn>15</mn><mtext>%</mtext></mrow></math></span>, and no cost at inference. Ablation studies confirmed that the combined curvature-divergence loss outperformed either component alone. A hyperparameter sweep revealed a broad optimal range, with folding artifacts monotonically decreasing as regularization strength increased, consistent with our theoretical proof.</div></div><div><h3>Conclusions:</h3><div>Enforcing second-order linear elasticity offers a robust and computationally efficient strategy to overcome the limitations of conventional regularizers. ElasticMorph provides a practical, principled, and plug-and-play solution for developing more accurate and physically plausible registration models, holding significant potential for improving the reliability of downstream biomedical image analysis tasks. The code will be made publicly available.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109182"},"PeriodicalIF":4.8,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145660631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The added value of radiomic analysis for predicting spontaneous preterm birth in the first trimester","authors":"William Cancino ,&nbsp;Carlos Hernan Becerra-Mojica ,&nbsp;Said Pertuz","doi":"10.1016/j.cmpb.2025.109181","DOIUrl":"10.1016/j.cmpb.2025.109181","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Preterm birth (PTB) is a public health problem. Researchers have worked to identify ways to detect women at risk for PTB early in pregnancy. Although existing biomarkers allow some women to be detected in the second trimester, detection in the first trimester is warranted to initiate earlier interventions. This study aims to explore the added value of radiomic analysis of transvaginal ultrasound (TVUS) images in the construction of first-trimester risk assessment models for predicting spontaneous preterm birth (sPTB).</div></div><div><h3>Methods:</h3><div>A retrospective cohort study was conducted including pregnant women who attended their screening ultrasound examination between 11+0 and 13+6 weeks. Data on medical history (MH), cervical length (CL), and cervical consistency index (CCI) were collected, along with TVUS images. These images were subjected to a computerized radiomic analysis to extract features, which were used to train different machine learning models incorporating a feature selection process. The area under the receiver operating characteristic curve (AUC) with 95% confidence intervals (CI) was used to evaluate the models’ performance in predicting sPTB.</div></div><div><h3>Results:</h3><div>A total of 253 pregnant women were included in the study, where 225 had a term birth and 28 had a sPTB. In sPTB prediction, MH, CL, and CCI obtained AUCs of 0.68 (95% CI, 0.56–0.79), 0.61 (95% CI, 0.48–0.73) and 0.67 (95% CI, 0.55–0.79), respectively. Among the evaluated machine learning models, logistic regression achieved the highest AUC and was therefore used to build the radiomic feature–based model (RF), which reached an AUC of 0.67 (95% CI, 0.56–0.76). When combining RF with MH and CCI, AUCs of 0.73 (95% CI, 0.63–0.82) and 0.74 (95% CI, 0.64–0.83) were achieved, respectively. The best performance was obtained by combining RF with both MH and CCI, reaching an AUC of 0.76 (95% CI, 0.68–0.84).</div></div><div><h3>Conclusion:</h3><div>The performance of the models improved by adding radiomic features, highlighting the potential of radiomic analysis for sPTB prediction in the first trimester.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109181"},"PeriodicalIF":4.8,"publicationDate":"2025-11-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145682518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adapt or specialize? A comprehensive evaluation of adapted SAM versus task-specific CNNs for fetal abdominal segmentation","authors":"Maria Chiara Fiorentino ,&nbsp;Lorenzo Federici ,&nbsp;Alessandro Pietro La Camera ,&nbsp;Enrico Gianluca Caiani","doi":"10.1016/j.cmpb.2025.109178","DOIUrl":"10.1016/j.cmpb.2025.109178","url":null,"abstract":"<div><h3>Background:</h3><div>The fetal abdomen is crucial in prenatal screening, offering key insights into fetal growth and congenital anomalies. However, segmenting internal abdominal structures in ultrasound (US) remains challenging due to anatomical variability, overlapping organs, and low contrast. While CNN-based models have shown strong performance in fetal head analysis, most existing methods focus on biometric measurements (e.g., head or abdominal circumference), leaving internal abdominal organ segmentation largely underexplored. Recently, foundation models like the Segment Anything Model (SAM) have emerged as flexible alternatives, enabling zero- or few-shot segmentation. Yet, their performance on fetal US remains poorly understood, and the need for adaptation is still an open question.</div></div><div><h3>Methods:</h3><div>We compare two segmentation strategies: (1) task-specific CNNs, including UNet, Attention UNet, nnUNet, DeepLabv3+, and their focal-loss variants; and (2) adapted SAM-based models. The latter includes zero-shot variants (SAMPoint, SAMBBox), pre-trained models (SAM Med2DBBox, MedSAMBBox), and adapted configurations, including lightweight fine-tuned models (SAM-LoRA, MedSAM-FrozenEncoder) and SAM Med2D variants with adapted layers. Experiments are conducted on a curated dataset of fetal abdominal US images with manual segmentations of the liver, stomach, artery, and umbilical vein. Performance is evaluated using Dice Similarity Coefficient, Intersection over Union, and precision. Statistical significance is assessed via pairwise Friedman chi-square tests.</div></div><div><h3>Results &amp; Conclusions:</h3><div>Zero-shot SAM variants performed poorly, particularly on small or low-contrast structures. In contrast, adapted SAM models consistently outperformed CNNs, reaching DSC scores up to 0.90 (liver) and 0.80 (artery). Prompt-based interaction enables semi-automated, human-in-the-loop workflows, supporting clinical applicability.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109178"},"PeriodicalIF":4.8,"publicationDate":"2025-11-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145600477","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An uncertainty-aware dynamic decision framework for progressive multi-omics integration in classification tasks","authors":"Nan Mu ,&nbsp;Hongbo Yang ,&nbsp;Chen Zhao","doi":"10.1016/j.cmpb.2025.109179","DOIUrl":"10.1016/j.cmpb.2025.109179","url":null,"abstract":"<div><h3>Background and Objective</h3><div>High-throughput multi-omics technologies offer significant potential for early disease diagnosis by capturing the complex interplay of molecular layers. However, two challenges hinder their practical application: First, single-omics approaches often miss the coordinated molecular interactions essential to disease understanding. Second, the high cost of comprehensive multi-omics profiling may lead to unnecessary resource use. To address these limitations, we propose an uncertainty-aware, multi-view dynamic decision framework that enhances classification accuracy while reducing testing costs.</div></div><div><h3>Methodology</h3><div>At the single-omics level, we incorporate subjective logic into neural network outputs, using refined activation functions to generate Dirichlet distribution parameters. This enables simultaneous estimation of belief masses (class-wise support) and uncertainty mass (prediction confidence). At the multi-omics level, we fuse complementary omics modalities via Dempster-Shafer theory to improve diagnostic accuracy and robustness. A dynamic decision mechanism incrementally integrates omics views for each patient, guided by real-time uncertainty: additional data are introduced only if prediction confidence fails to meet a predefined threshold, enabling individualized and efficient decision-making.</div></div><div><h3>Results and Conclusion</h3><div>We evaluate our approach on four benchmark multi-omics datasets, i.e., ROSMAP, LGG, BRCA, and KIPAN. In three datasets, over 50 % of cases achieved accurate classification using a single omics modality, effectively reducing redundant testing. Meanwhile, our method maintains diagnostic performance comparable to full-omics models and preserves essential biological insights. This framework introduces a novel 'on-demand testing' paradigm into precision medicine, enabling intelligent resource allocation and reducing healthcare costs, which is especially valuable in resource-limited clinical environments.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109179"},"PeriodicalIF":4.8,"publicationDate":"2025-11-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145647488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MDM-DTA: Message Passing Neural Network with molecular descriptors and Mixture of Experts for drug–target affinity prediction","authors":"Yang Dai ,&nbsp;Xiaoyu Tan ,&nbsp;Haoyu Wang ,&nbsp;Gengchen Ma ,&nbsp;Yujie Xiong ,&nbsp;Xihe Qiu","doi":"10.1016/j.cmpb.2025.109163","DOIUrl":"10.1016/j.cmpb.2025.109163","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Drug–target affinity (DTA) prediction is a pivotal task in computational drug discovery, enabling the estimation of binding affinities between small molecules and their target proteins. This process is essential for reducing the costs, development time, and risks inherent in traditional drug development pipelines. Current DTA prediction models primarily rely on separate extraction and concatenation of drug and protein features. However, these models often fail to account for the complex semantic relationships within protein sequences, which limits their ability to accurately predict affinity.</div></div><div><h3>Methods:</h3><div>In response to these challenges, we propose MDM-DTA, a novel framework leveraging a Mixture of Experts (MoE) strategy to integrate diverse molecular and protein representations. For drug representation, MDM-DTA utilizes molecular graphs, which are processed via Message Passing Neural Networks (MPNNs), alongside molecular descriptors that are passed through a three-layer convolutional neural network (CNN). Protein features are extracted using a deep convolutional network enhanced with Squeeze-and-Excitation (SE) mechanisms to capture inter-channel dependencies. Furthermore, protein sequence semantics are encoded through pre-trained embeddings from a knowledge-guided Bidirectional Encoder Representations from Transformers (BERT) model and the Evolutionary Scale Modeling 2 (ESM2) model, enabling the model to capture contextual relationships within protein sequences.</div></div><div><h3>Results:</h3><div>Extensive experiments on three benchmark datasets demonstrate that MDM-DTA consistently outperforms state-of-the-art models of similar complexity in terms of predictive accuracy. The incorporation of both structural and semantic features significantly enhances the model’s ability to predict drug–target binding affinities, highlighting the importance of a multi-modal representation approach.</div></div><div><h3>Conclusions:</h3><div>The proposed MDM-DTA framework effectively integrates both molecular and semantic protein representations, providing superior performance in DTA prediction tasks. The results underscore the potential of MDM-DTA to improve the accuracy of computational drug discovery models, facilitating the identification of novel drug candidates and advancing the field of in silico drug development.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"274 ","pages":"Article 109163"},"PeriodicalIF":4.8,"publicationDate":"2025-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145596058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Novel machine learning framework for multidimensional biological age estimation reveals heterogeneous aging of organ systems","authors":"Qi Yu ,&nbsp;Lijuan Da ,&nbsp;Qian Ma ,&nbsp;Yushu Huang ,&nbsp;Yue Dong ,&nbsp;Yuan Liu ,&nbsp;Xiaoyu Li ,&nbsp;Xifeng Wu ,&nbsp;Zilin Li ,&nbsp;Wenyuan Li","doi":"10.1016/j.cmpb.2025.109176","DOIUrl":"10.1016/j.cmpb.2025.109176","url":null,"abstract":"<div><h3>Objective</h3><div>Existing biological age (BA) models often oversimplify aging’s complexity, offering single-dimensional metrics. However, these fail to capture the critical heterogeneity of aging across organs.. This study aims to develop a machine learning-based unified framework to assess and interpret multi-organ biological aging comprehensively.</div></div><div><h3>Method</h3><div>Using data from UK Biobank participants, we trained and integrated organ-specific BA estimates to assess multidimensional BA within an ensemble learning framework, and uncover distinct aging patterns.</div></div><div><h3>Result</h3><div>Our Fusion BA (an overall estimation of BA) was significantly correlated with chronological age (CA) (mean absolute error (MAE): 4.473 years; Pearson correlation: 0.718, <em>P</em> &lt; 0.01). Accelerated Fusion BA derived from the ensemble model (contrast between Fusion BA and CA) predicted 10-year mortality (HR=1.504, 95 % CI: 1.438–1.574). Organ-specific BA correlated with organ disease risk and effectively captured distinct aging patterns.</div></div><div><h3>Conclusion</h3><div>This framework enables systemic and organ-specific aging assessment, provides actionable tools and insights for clinical risk.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"275 ","pages":"Article 109176"},"PeriodicalIF":4.8,"publicationDate":"2025-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145682465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Drug repurposing through pathway perturbation dynamics: A systems biology approach for precision oncology","authors":"Xianbin Li ,&nbsp;Wuxiang Ruan ,&nbsp;Guoan Lu ,&nbsp;Dingcheng Ban ,&nbsp;Luming Tian ,&nbsp;Binbin Wang ,&nbsp;Tao Liu ,&nbsp;Guodao Zhang ,&nbsp;Chunping Wang ,&nbsp;Jie Lin","doi":"10.1016/j.cmpb.2025.109177","DOIUrl":"10.1016/j.cmpb.2025.109177","url":null,"abstract":"<div><h3>Background and objective</h3><div>Drug repurposing offers a cost-efficient strategy to discover new therapeutic applications for approved drugs. While current computational strategies prioritize candidates by targeting disease-related pathways, they often fail to quantitatively model pathway perturbation dynamics—a critical gap that limits mechanistic interpretability.</div></div><div><h3>Methods</h3><div>To address this issue, we presented PathPertDrug, a novel framework that systematically identifies cancer drug candidates by quantifying functional antagonism between drug-induced and disease-associated pathway perturbations (activation/ inhibition). By integrating drug-induced gene expression, disease-related gene expression, and pathway information, PathPertDrug evaluated pathway-level functional reversals, enabling precise prediction of drug-disease associations.</div></div><div><h3>Results</h3><div>Our method demonstrated superior predictive accuracy and robustness across pan-cancer benchmarks. It achieved a higher median AUROC (0.62 vs. 0.42–0.53) and a substantial improvement in AUPR (3–23 %) over existing methods. The consistent AUPR enhancement, particularly under class imbalance, underscores the robustness of our model in reliably prioritizing true positive associations. Validated by the comparative toxicogenmics database, PathPertDrug rediscovered 83 % of literature-supported cancer drugs (e.g., fulvestrant (Fulvestrant) for colorectal cancer) and predicted novel candidates (e.g., rifabutin–lung cancer).</div></div><div><h3>Conclusions</h3><div>This pathway-centric approach bridged mechanistic insights with translational applications, providing a paradigm shift for precision oncology drug discovery.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"274 ","pages":"Article 109177"},"PeriodicalIF":4.8,"publicationDate":"2025-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145602799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-constituent model of thrombosis for blood-contacting medical devices","authors":"Yuning Lin ,&nbsp;Yuzhou Cheng ,&nbsp;Kaiyue Yang ,&nbsp;Kun Luo ,&nbsp;Jianren Fan ,&nbsp;Ru Lin ,&nbsp;Qiang Shu","doi":"10.1016/j.cmpb.2025.109158","DOIUrl":"10.1016/j.cmpb.2025.109158","url":null,"abstract":"<div><h3>Background and objectives:</h3><div>Ventricular assist devices (VADs) are currently effective clinical interventions for treating diseases related to end-stage heart failure (HF), yet hemocompatibility-related complications persist as prevalent challenges. Among them, device-induced thrombosis may lead to severe consequences such as stroke, neurological events, pump replacement, and even mortality, making it critically important to predict thrombosis accurately. The primary objective of this study is to develop a thrombosis model capable of simulating thrombus formation within VADs.</div></div><div><h3>Methods:</h3><div>The proposed model integrates hemodynamics, platelet activity, and the coagulation cascade, wherein the cascade products regulate platelet activation, aggregation, and stabilization. To enable simulations at the device scale while preserving essential physiological mechanisms, a reduced-order coagulation cascade model was adopted. Furthermore, the model incorporates hemodynamic-thrombus interaction while thrombus breakdown due to the shear stress clearance is under consideration.</div></div><div><h3>Results:</h3><div>The model was first validated in a backward-facing step (BFS) geometry to assess its applicability under separated flow conditions. In terms of volumetric evolution, the simulation followed a trend consistent with experimental data, while the thrombus length and height matched the experimental measurements closely. The model was then applied to a left ventricular assist device (VAD) to explore thrombus formation mechanisms. Simulations at different flow rates revealed consistent thrombosis on the straightener blades, while initiation sites and growth dynamics were governed by local hemodynamics, platelet activation, and stabilization.</div></div><div><h3>Conclusions:</h3><div>The thrombosis model developed in this study enables the investigation of thrombus formation mechanisms and the identification of potential high-risk regions within VADs under varying flow conditions. It provides a basis for future experimental validation and has potential utility for optimizing VAD design and informing patient-specific risk assessment.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"274 ","pages":"Article 109158"},"PeriodicalIF":4.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145602764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GICAF-Net: A cross-attentional graph-image fusion network for hyperspectral pathological diagnosis of FNH and HCC","authors":"Yunze Li ,&nbsp;Haiyan Chen ,&nbsp;Baoxian Gong ,&nbsp;Jiankang Han ,&nbsp;Jun Cheng ,&nbsp;Shuai Gao ,&nbsp;Wei Li","doi":"10.1016/j.cmpb.2025.109171","DOIUrl":"10.1016/j.cmpb.2025.109171","url":null,"abstract":"<div><h3>Background and objective</h3><div>Accurate intraoperative differentiation between focal nodular hyperplasia (FNH) and hepatocellular carcinoma (HCC) remains a major clinical challenge, especially in atypical cases where conventional imaging and histopathology are constrained by turnaround time, cost, or spectral resolution. This study aims to develop a novel deep learning framework to improve the precision and efficiency of hyperspectral pathological diagnosis for liver tumors.</div></div><div><h3>Methods</h3><div>We propose GICAF-Net, a Graph–Image Cross-Attentional Fusion Network, designed to leverage hyperspectral imaging (HSI) for capturing fine-grained spatial–spectral information. The network employs a dual-branch architecture: (1) a residual convolutional branch for extracting pseudo-color image features, and (2) a residual graph convolutional branch for modeling topological spatial–spectral features. A Topology-Aware Cross-Attention Fusion (TACA) module enables bidirectional information exchange between the two modalities, while a multi-constraint fusion loss—combining cross-entropy, prediction confidence, and cross-modal attention consistency—enhances classification stability. A balanced hyperspectral liver tumor dataset consisting of 60 HCC and 60 FNH cases was constructed and evaluated using ten-fold cross-validation.</div></div><div><h3>Results</h3><div>GICAF-Net achieved an AUC of 0.9571 ± 0.0068, accuracy of 88.34 % ± 1.10 %, and F1-score of 88.32 % ± 1.11 %, outperforming state-of-the-art baseline models. Ablation experiments further validated the contributions of both the TACA module and the multi-constraint loss function in enhancing cross-modal fusion and improving classification performance.</div></div><div><h3>Conclusion</h3><div>The integration of graph-based spectral–structural modeling with deep visual features through cross-attention provides a powerful approach for hyperspectral pathological diagnosis. The proposed GICAF-Net demonstrates strong potential for rapid, accurate, and minimally invasive intraoperative differentiation of FNH and HCC, offering valuable clinical support in liver tumor surgery.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"274 ","pages":"Article 109171"},"PeriodicalIF":4.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145602891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}