Computers in biology and medicine最新文献_第7页

Mixed-effects additive Bayesian networks for the assessment of ruptured intracranial aneurysms: Insights from multicenter data 用于颅内动脉瘤破裂评估的混合效应加性贝叶斯网络：来自多中心数据的见解。

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-23 DOI: 10.1016/j.compbiomed.2025.111380

Matteo Delucchi , Philippe Bijlenga , Sandrine Morel , Reinhard Furrer , Isabel C. Hostettler , Mark K. Bakker , Romain Bourcier , Antti Lindgren , Svenja Maschke , Oliver Bozinov , Henry Houlden , David Werring , Ynte M. Ruigrok , Maria Wostrack , Bernhard Meyer , Marian Christoph Neidert , International Stroke Genetics Consortium (ISGC) Intracranial Aneurysm Working Group , Genetics and Observational Subarachnoid Haemorrhage (GOSH) Study Investigators , Sven Hirsch , Georg R. Spinner

Background

Multicenter clinical datasets are increasingly available but are often analyzed using models that ignore center-specific variability or fail to capture complex dependencies among variables. This is particularly limiting in the context of intracranial aneurysms (IAs), where existing risk scores lack precision and interpretability.

Objective

We apply mixed-effects additive Bayesian networks (ABNs) to a large multicenter dataset to model interdependencies among demographic, clinical, and aneurysm-specific features in solitary ruptured IAs, while accounting for heterogeneity across study centers.

Methods

Data from seven European centers were subsequently harmonized into an observational, cross-sectional cohort of patients with solitary ruptured IAs to mitigate selection bias inherent in cohorts with incidentally discovered unruptured IAs. Mixed-effects ABNs modeled probabilistic dependencies using generalized linear models with random intercepts for study centers. Structure learning was performed via structural Markov chain Monte Carlo sampling with Bayesian information criterion scoring. Model performance was assessed using graphical comparison, intraclass correlation coefficients (ICCs), and predictive metrics.

Results

The mixed-effects ABN produced a more parsimonious network than the pooled model and better captured center-specific variation, particularly for variables with high ICCs (e.g., family history: ICC = 0.369). It also outperformed the pooled ABN in predicting family history (Area Under the Curve = 0.694 vs. 0.585) while yielding clinically interpretable associations, such as the influence of sex, smoking, and IA location on IA size.

Conclusion

This application of mixed-effects ABNs reveals that accounting for inter-center heterogeneity is critical for accurately modeling risk factor dependencies in multicenter IA cohorts. This approach yields a more parsimonious network structure by reducing spurious associations found in pooled models. By disentangling patient-level effects from center-specific variations, the model enhances predictive power for heterogeneous variables and provides more reliable, clinically interpretable insights into IA pathophysiology, advancing the potential for personalized risk assessment.

背景：多中心临床数据集越来越多，但通常使用忽略中心特异性变异性或无法捕获变量之间复杂依赖关系的模型进行分析。这在颅内动脉瘤（IAs）的情况下尤其有限，因为现有的风险评分缺乏准确性和可解释性。目的：我们将混合效应加性贝叶斯网络（ABNs）应用于一个大型多中心数据集，以模拟孤立性破裂IAs中人口统计学、临床和动脉瘤特异性特征之间的相互依赖性，同时考虑研究中心之间的异质性。方法：来自7个欧洲中心的数据随后被统一到一个观察性的横断面队列中，该队列中有孤立性破裂的IAs患者，以减轻偶然发现的未破裂IAs队列固有的选择偏差。混合效应ABNs使用具有随机截距的广义线性模型对研究中心的概率依赖性进行建模。结构学习是通过结构马尔可夫链蒙特卡罗采样和贝叶斯信息准则评分来完成的。通过图形比较、类内相关系数（ICCs）和预测指标评估模型性能。结果：与混合模型相比，混合效应ABN产生了一个更简洁的网络，并更好地捕获了中心特异性变化，特别是对于ICC高的变量（例如，家族史：ICC = 0.369）。它在预测家族史（曲线下面积= 0.694 vs. 0.585）方面也优于合并ABN，同时产生临床可解释的关联，如性别、吸烟和IA位置对IA大小的影响。结论：混合效应ABNs的应用表明，在多中心IA队列中，考虑中心间异质性对于准确建模风险因素依赖关系至关重要。这种方法通过减少在池化模型中发现的虚假关联，产生更简洁的网络结构。通过将患者层面的影响与中心特异性变化分离开来，该模型增强了对异质变量的预测能力，并为IA病理生理学提供了更可靠、临床可解释的见解，提高了个性化风险评估的潜力。

{"title":"Mixed-effects additive Bayesian networks for the assessment of ruptured intracranial aneurysms: Insights from multicenter data","authors":"Matteo Delucchi , Philippe Bijlenga , Sandrine Morel , Reinhard Furrer , Isabel C. Hostettler , Mark K. Bakker , Romain Bourcier , Antti Lindgren , Svenja Maschke , Oliver Bozinov , Henry Houlden , David Werring , Ynte M. Ruigrok , Maria Wostrack , Bernhard Meyer , Marian Christoph Neidert , International Stroke Genetics Consortium (ISGC) Intracranial Aneurysm Working Group , Genetics and Observational Subarachnoid Haemorrhage (GOSH) Study Investigators , Sven Hirsch , Georg R. Spinner","doi":"10.1016/j.compbiomed.2025.111380","DOIUrl":"10.1016/j.compbiomed.2025.111380","url":null,"abstract":"<div><h3>Background</h3><div>Multicenter clinical datasets are increasingly available but are often analyzed using models that ignore center-specific variability or fail to capture complex dependencies among variables. This is particularly limiting in the context of intracranial aneurysms (IAs), where existing risk scores lack precision and interpretability.</div></div><div><h3>Objective</h3><div>We apply mixed-effects additive Bayesian networks (ABNs) to a large multicenter dataset to model interdependencies among demographic, clinical, and aneurysm-specific features in solitary ruptured IAs, while accounting for heterogeneity across study centers.</div></div><div><h3>Methods</h3><div>Data from seven European centers were subsequently harmonized into an observational, cross-sectional cohort of patients with solitary ruptured IAs to mitigate selection bias inherent in cohorts with incidentally discovered unruptured IAs. Mixed-effects ABNs modeled probabilistic dependencies using generalized linear models with random intercepts for study centers. Structure learning was performed via structural Markov chain Monte Carlo sampling with Bayesian information criterion scoring. Model performance was assessed using graphical comparison, intraclass correlation coefficients (ICCs), and predictive metrics.</div></div><div><h3>Results</h3><div>The mixed-effects ABN produced a more parsimonious network than the pooled model and better captured center-specific variation, particularly for variables with high ICCs (e.g., family history: ICC = 0.369). It also outperformed the pooled ABN in predicting family history (Area Under the Curve = 0.694 vs. 0.585) while yielding clinically interpretable associations, such as the influence of sex, smoking, and IA location on IA size.</div></div><div><h3>Conclusion</h3><div>This application of mixed-effects ABNs reveals that accounting for inter-center heterogeneity is critical for accurately modeling risk factor dependencies in multicenter IA cohorts. This approach yields a more parsimonious network structure by reducing spurious associations found in pooled models. By disentangling patient-level effects from center-specific variations, the model enhances predictive power for heterogeneous variables and provides more reliable, clinically interpretable insights into IA pathophysiology, advancing the potential for personalized risk assessment.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111380"},"PeriodicalIF":6.3,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145827079","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}

引用次数: 0

Innovative virtual reality application based on proprioceptive ‘facilitation and inhibition’ to improve upper limb function: A prospective, feasibility and proof-of-concept study in people with multiple sclerosis 基于本体感觉“促进和抑制”的创新虚拟现实应用改善上肢功能：多发性硬化症患者的前瞻性、可行性和概念验证研究

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-23 DOI: 10.1016/j.compbiomed.2025.111376

Jakub Frank , Libor Váša , Guillaume Lavoué , Barbora Miznerová , Lubomír Rodina , Jindra Reissigová , Anna Herynková , Petra Reckziegelová , Kamila Řasová

We present a comprehensive software system developed for an innovative virtual reality application that leverages the principles of proprioceptive ‘facilitation and inhibition’ to enhance upper limb function, along with the results of a prospective, feasibility and proof-of-concept study conducted in people with multiple sclerosis. The system tracks upper limb movement at two locations, as well as the movement of the chest and head. Each tracked point continuously provides real-time spatial position and orientation data to the system. Participants engage in physical therapy based on the principles of proprioceptive ‘facilitation and inhibition’, interspersed with selected relaxation games, while their performance — relative to predefined exercise templates — is continuously monitored. All session data is also stored for retrospective analysis. The system delivers real-time performance evaluation to the therapist, along with live feedback and stereoscopic 3D guidance for the participant, thereby enhancing engagement in movement. Preliminary measurements indicate promising results, particularly in terms of improved gross motor function compared to the same physiotherapy without the virtual reality component.

我们提出了一个综合的软件系统，开发了一个创新的虚拟现实应用程序，利用本体感觉的“促进和抑制”的原则来增强上肢功能，以及在多发性硬化症患者中进行的前瞻性、可行性和概念验证研究的结果。该系统跟踪上肢的两个位置的运动，以及胸部和头部的运动。每个跟踪点连续不断地向系统提供实时的空间位置和方向数据。参与者根据本体感觉的“促进和抑制”原则进行物理治疗，其间穿插选定的放松游戏，同时他们的表现（相对于预定义的运动模板）被持续监测。所有的会话数据也被存储以进行回顾性分析。该系统为治疗师提供实时性能评估，同时为参与者提供实时反馈和立体3D指导，从而提高运动的参与度。初步测量结果表明，与没有虚拟现实组件的相同物理治疗相比，特别是在改善大运动功能方面。

{"title":"Innovative virtual reality application based on proprioceptive ‘facilitation and inhibition’ to improve upper limb function: A prospective, feasibility and proof-of-concept study in people with multiple sclerosis","authors":"Jakub Frank , Libor Váša , Guillaume Lavoué , Barbora Miznerová , Lubomír Rodina , Jindra Reissigová , Anna Herynková , Petra Reckziegelová , Kamila Řasová","doi":"10.1016/j.compbiomed.2025.111376","DOIUrl":"10.1016/j.compbiomed.2025.111376","url":null,"abstract":"<div><div>We present a comprehensive software system developed for an innovative virtual reality application that leverages the principles of proprioceptive ‘facilitation and inhibition’ to enhance upper limb function, along with the results of a prospective, feasibility and proof-of-concept study conducted in people with multiple sclerosis. The system tracks upper limb movement at two locations, as well as the movement of the chest and head. Each tracked point continuously provides real-time spatial position and orientation data to the system. Participants engage in physical therapy based on the principles of proprioceptive ‘facilitation and inhibition’, interspersed with selected relaxation games, while their performance — relative to predefined exercise templates — is continuously monitored. All session data is also stored for retrospective analysis. The system delivers real-time performance evaluation to the therapist, along with live feedback and stereoscopic 3D guidance for the participant, thereby enhancing engagement in movement. Preliminary measurements indicate promising results, particularly in terms of improved gross motor function compared to the same physiotherapy without the virtual reality component.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111376"},"PeriodicalIF":6.3,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145827131","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}

引用次数: 0

MLDeCNV: A machine learning approach for predicting copy number variation types in plant genomes MLDeCNV：一种预测植物基因组拷贝数变异类型的机器学习方法。

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-22 DOI: 10.1016/j.compbiomed.2025.111394

Parinita Das , Bibek Saha , Nitesh Kumar Sharma , Mir Asif Iquebal , Alexie Papanicolaou , U.B. Angadi , Dinesh Kumar , Sarika Jaiswal

Copy number variations (CNVs) play a crucial role in shaping genetic diversity and influencing various plant traits. However, existing methods for CNV characterization often face challenges due to the complexity and repetitive nature of plant genomes. Here, we present MLDeCNV (Machine Learning for Decoding Copy Number Variation) a novel open-source machine-learning based tool optimized for predicting CNV types (deletions, duplications, and non-CNVs) in plant genomes. Built on the XGBoost model, MLDeCNV utilizes 32 selected CNV-related features derived from coverage metrics, nucleotide composition, and sequencing statistics. The model was trained on a high-confidence CNV dataset comprising of experimentally validated and computationally predicted CNVs. It exhibits strong performance across various CNV size ranges and training set sizes, achieving an accuracy of 89.27 %, with precision, recall, and F1-score, all at 89.3 %, and an Area Under Curve of 0.9783, underscoring its robustness and reliability. Extensive comparisons with traditional machine learning models reveal that XGBoost outperforms other methods, particularly in handling complex, nonlinear interactions within the CNV data. Additionally, while MLDeCNV does not perform de novo CNV detection, it evaluates CNV type classification from pre-identified genomic regions, making it a post-detection classification tool. This tool, accessible at http://46.202.167.198:5004/ can be integrated downstream of CNV detection pipelines, enhancing the accuracy of CNV type categorization. The precise classification of CNV types from pre-identified genomic regions will streamline downstream genomic analyses, facilitating enhanced understanding and utilization of genetic variation in plants.

拷贝数变异（Copy number variation, CNVs）在形成遗传多样性和影响植物各种性状方面起着至关重要的作用。然而，由于植物基因组的复杂性和重复性，现有的CNV鉴定方法经常面临挑战。在这里，我们提出了MLDeCNV（解码拷贝数变化的机器学习），这是一种新的基于开源机器学习的工具，用于预测植物基因组中的CNV类型（缺失、重复和非CNV）。MLDeCNV建立在XGBoost模型的基础上，利用32个选择的cnv相关特征，这些特征来自覆盖度量、核苷酸组成和测序统计。该模型在高置信度CNV数据集上进行训练，该数据集包括实验验证和计算预测的CNV。它在不同的CNV大小范围和训练集大小上都表现出很强的性能，准确率为89.27%，精密度、召回率和f1得分均为89.3%，曲线下面积为0.9783，表明了它的鲁棒性和可靠性。与传统机器学习模型的广泛比较表明，XGBoost优于其他方法，特别是在处理CNV数据中复杂的非线性相互作用方面。此外，虽然MLDeCNV不执行从头CNV检测，但它根据预先识别的基因组区域评估CNV类型分类，使其成为检测后分类工具。该工具可在http://46.202.167.198:5004/上访问，可以集成在CNV检测管道的下游，提高CNV类型分类的准确性。从预先鉴定的基因组区域精确分类CNV类型将简化下游基因组分析，促进对植物遗传变异的理解和利用。

{"title":"MLDeCNV: A machine learning approach for predicting copy number variation types in plant genomes","authors":"Parinita Das , Bibek Saha , Nitesh Kumar Sharma , Mir Asif Iquebal , Alexie Papanicolaou , U.B. Angadi , Dinesh Kumar , Sarika Jaiswal","doi":"10.1016/j.compbiomed.2025.111394","DOIUrl":"10.1016/j.compbiomed.2025.111394","url":null,"abstract":"<div><div>Copy number variations (CNVs) play a crucial role in shaping genetic diversity and influencing various plant traits. However, existing methods for CNV characterization often face challenges due to the complexity and repetitive nature of plant genomes. Here, we present <em>MLDeCNV</em> (Machine Learning for Decoding Copy Number Variation) a novel open-source machine-learning based tool optimized for predicting CNV types (deletions, duplications, and non-CNVs) in plant genomes. Built on the XGBoost model, <em>MLDeCNV</em> utilizes 32 selected CNV-related features derived from coverage metrics, nucleotide composition, and sequencing statistics. The model was trained on a high-confidence CNV dataset comprising of experimentally validated and computationally predicted CNVs. It exhibits strong performance across various CNV size ranges and training set sizes, achieving an accuracy of 89.27 %, with precision, recall, and F1-score, all at 89.3 %, and an Area Under Curve of 0.9783, underscoring its robustness and reliability. Extensive comparisons with traditional machine learning models reveal that XGBoost outperforms other methods, particularly in handling complex, nonlinear interactions within the CNV data. Additionally, while <em>MLDeCNV</em> does not perform <em>de novo</em> CNV detection, it evaluates CNV type classification from pre-identified genomic regions, making it a post-detection classification tool. This tool, accessible at <span><span>http://46.202.167.198:5004/</span><svg><path></path></svg></span> can be integrated downstream of CNV detection pipelines, enhancing the accuracy of CNV type categorization. The precise classification of CNV types from pre-identified genomic regions will streamline downstream genomic analyses, facilitating enhanced understanding and utilization of genetic variation in plants.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111394"},"PeriodicalIF":6.3,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145818349","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}

引用次数: 0

Patient-derived geometry-integrated multiphysics framework: Computational optimization of immunotherapeutic nanoparticle delivery via tumor microenvironment reprogramming in prostate cancer 患者衍生的几何集成多物理框架：通过前列腺癌肿瘤微环境重编程的免疫治疗纳米颗粒递送的计算优化

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-20 DOI: 10.1016/j.compbiomed.2025.111388

Fatemeh Mirala , M. Soltani

The immune checkpoint inhibitor (ICB) resistance of prostate cancer is due to its peculiar tumor microenvironment (TME), dense collagen stromal and disorganized vascularity, increased interstitial fluid pressure (IFP), blocked drugs and also immune cell penetration, and fostered immunosuppression. None of the generic computational models considered all these factors together in their modelin before. To address these problems, we developed a patient-specific multiphysics framework that integrates three-dimensional (3D) magnetic resonance imaging (MRI)-derived tumor geometries of prostate from the Cancer Imaging Archive (TCIA), TME normalization procedures (vascular and stromal), nanocarrier-based PD-1/anti-PD-L1 delivery, with the simulation time extended to 50 days, and long-term responses due to immune memory or recurrence, beyond those observed in the previously described less than 30-day simulations.

Vascular normalization optimized vessel permeability and functional density, while stromal normalization reduced extracellular matrix (ECM) stiffness, effectively reprogramming the TME across all patient tumors. In this simulation, the value of the IFP reduced by over 70 % (from 1458–1484 Pa to 438–445 Pa), and the interstitial fluid velocity (IFV) increased by 2–3 folds, with a mechanical stress reduction of 42 %. This permissive environment facilitated the tumor penetration of nano-ICB, allowing therapeutic concentration levels to be reached in the tumor interstitium (from 0.0039 nmol/ml to 0.0058 nmol/ml). Immune system optimization, involving increasing cytotoxic T cell (CD8⁺ T cell) entry (from 150 day⁻¹ to 300 day⁻¹) and regulatory T cells (Treg) rate of death (from 0.02 day⁻¹ to 0.05 day⁻¹), drove the fraction of killed cancer cells (FKCs) to 0.52 in Tumor 3 and 0.84 in Tumor 2. Model validation confirmed numerical robustness via mesh (184,254 elements) and time-step (0.1 min) independence, with less than 2 % variation in key outputs (IFP, IFV, tumor volume, and FKC).

The presented study moves the frontiers of computational oncology by relating the anatomical specifics of the patient to the mechanics of the TME, on one hand, and nano-immunotherapy, on the other, advancing beyond the simplifications of common geometric representations and one-component models. It provides a clinically applicable platform for the optimization of individualized treatment of prostate cancer based on the prediction of the effectiveness of synergies between TME normalization and nano-ICB to enhance anti-tumor efficacy.

前列腺癌的免疫检查点抑制剂（ICB）耐药是由于其特殊的肿瘤微环境（TME）、致密的胶原基质和紊乱的血管、间质液压力（IFP）升高、阻断药物和免疫细胞渗透以及促进免疫抑制。以前没有一个通用的计算模型在它们的模型中一起考虑所有这些因素。为了解决这些问题，我们开发了一个针对患者的多物理场框架，该框架集成了来自癌症成像档案（TCIA）的三维（3D）磁共振成像（MRI）衍生的前列腺肿瘤几何形状、TME正常化程序（血管和基质）、基于纳米载体的PD-1/抗pd - l1递送，模拟时间延长至50天，以及由于免疫记忆或复发引起的长期反应。除了在之前描述的不到30天的模拟中观察到的那些。血管正常化优化了血管通透性和功能密度，而间质正常化降低了细胞外基质（ECM）硬度，有效地重新编程了所有患者肿瘤的TME。在此模拟中，IFP值降低了70%以上（从1458-1484 Pa降至438-445 Pa），间质流体速度（IFV）增加了2-3倍，机械应力降低了42%。这种宽松的环境促进了纳米icb的肿瘤渗透，使其在肿瘤间质中达到治疗浓度水平（从0.0039 nmol/ml到0.0058 nmol/ml）。免疫系统优化，包括增加细胞毒性T细胞（CD8+ T细胞）进入（从150天−1增加到300天−1）和调节性T细胞（Treg）死亡率（从0.02天−1增加到0.05天−1），使肿瘤3中死亡癌细胞（FKCs）的比例达到0.52，肿瘤2中达到0.84。模型验证通过网格（184,254个元素）和时间步（0.1分钟）独立性证实了数值稳健性，关键输出（IFP， IFV，肿瘤体积和FKC）的变化小于2%。所提出的研究通过将患者的解剖细节与TME的力学联系起来，以及将纳米免疫疗法与TME力学联系起来，推动了计算肿瘤学的前沿，超越了常见几何表示和单组分模型的简化。通过预测TME归一化与纳米icb协同增强抗肿瘤疗效的有效性，为优化前列腺癌个体化治疗提供了临床应用平台。

{"title":"Patient-derived geometry-integrated multiphysics framework: Computational optimization of immunotherapeutic nanoparticle delivery via tumor microenvironment reprogramming in prostate cancer","authors":"Fatemeh Mirala , M. Soltani","doi":"10.1016/j.compbiomed.2025.111388","DOIUrl":"10.1016/j.compbiomed.2025.111388","url":null,"abstract":"<div><div>The immune checkpoint inhibitor (ICB) resistance of prostate cancer is due to its peculiar tumor microenvironment (TME), dense collagen stromal and disorganized vascularity, increased interstitial fluid pressure (IFP), blocked drugs and also immune cell penetration, and fostered immunosuppression. None of the generic computational models considered all these factors together in their modelin before. To address these problems, we developed a patient-specific multiphysics framework that integrates three-dimensional (3D) magnetic resonance imaging (MRI)-derived tumor geometries of prostate from the Cancer Imaging Archive (TCIA), TME normalization procedures (vascular and stromal), nanocarrier-based PD-1/anti-PD-L1 delivery, with the simulation time extended to 50 days, and long-term responses due to immune memory or recurrence, beyond those observed in the previously described less than 30-day simulations.</div><div>Vascular normalization optimized vessel permeability and functional density, while stromal normalization reduced extracellular matrix (ECM) stiffness, effectively reprogramming the TME across all patient tumors. In this simulation, the value of the IFP reduced by over 70 % (from 1458–1484 Pa to 438–445 Pa), and the interstitial fluid velocity (IFV) increased by 2–3 folds, with a mechanical stress reduction of 42 %. This permissive environment facilitated the tumor penetration of nano-ICB, allowing therapeutic concentration levels to be reached in the tumor interstitium (from 0.0039 nmol/ml to 0.0058 nmol/ml). Immune system optimization, involving increasing cytotoxic T cell (CD8<sup>+</sup> T cell) entry (from 150 day<sup>−1</sup> to 300 day<sup>−1</sup>) and regulatory T cells (Treg) rate of death (from 0.02 day<sup>−1</sup> to 0.05 day<sup>−1</sup>), drove the fraction of killed cancer cells (FKCs) to 0.52 in Tumor 3 and 0.84 in Tumor 2. Model validation confirmed numerical robustness via mesh (184,254 elements) and time-step (0.1 min) independence, with less than 2 % variation in key outputs (IFP, IFV, tumor volume, and FKC).</div><div>The presented study moves the frontiers of computational oncology by relating the anatomical specifics of the patient to the mechanics of the TME, on one hand, and nano-immunotherapy, on the other, advancing beyond the simplifications of common geometric representations and one-component models. It provides a clinically applicable platform for the optimization of individualized treatment of prostate cancer based on the prediction of the effectiveness of synergies between TME normalization and nano-ICB to enhance anti-tumor efficacy.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111388"},"PeriodicalIF":6.3,"publicationDate":"2025-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145799568","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}

引用次数: 0

Exploring phytochemical inhibitors of fatty acid elongase ELOVL6 for targeted treatment of chronic myeloid leukemia: A comprehensive network-based drug discovery approach 探索脂肪酸延长酶ELOVL6的植物化学抑制剂靶向治疗慢性髓性白血病：一种基于网络的综合药物发现方法

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-19 DOI: 10.1016/j.compbiomed.2025.111342

Alvea Tasneem , Manish Singh , Hridoy R. Bairagya , Ravins Dohare

Quiescent leukemic stem cells (LSCs) that persist in the bone marrow microenvironment are responsible for chronic myeloid leukemia (CML) relapses and tyrosine kinase inhibitors (TKIs) resistance. This highlights a critical need to uncover alternative gene targets and pathways involved in LSC maintenance. Network biology in drug development has become essential for predicting drug targets in CML disease. This present computational study aims to identify key regulatory genes that are differentially expressed and involved in molecular pathway alternative to BCR-ABL, which may facilitate the eradication of leukemic stem and progenitor cells. Comparative analysis between CML stem and progenitor cells and their normal counterparts revealed 182 differentially expressed genes (DEGs). Applying Weighted Gene Co-expression Network Algorithm (WGCNA) identified a significant gene module comprising 73 hub genes. Protein-protein interaction and enrichment analyses indicated these genes are involved in mitochondrial translation elongation, steroid metabolism, cholesterol, and fatty acyl-CoA biosynthesis. Furthermore, a three-node regulatory network composed of hub genes, CML-associated transcription factors (TFs), and differentially expressed microRNAs (DEMs) was constructed, highlighting three key regulators: ELOVL6, SP1 (TF), and miR-1207-5p. To explore the therapeutic potential of the overexpressed target gene ELOVL6, we performed high-throughput virtual screening of phytochemical compounds against the ELOVL6 protein structure. Subsequent molecular docking, pharmacokinetics, toxicity, and molecular dynamics (MD) simulations revealed two phytochemicals — withaphysalin A and chelidimerine —as potential inhibitors of the ELOVL6 therapeutic biomarker in CML.

在骨髓微环境中持续存在的静止白血病干细胞（LSCs）是慢性髓性白血病（CML）复发和酪氨酸激酶抑制剂（TKIs）耐药的原因。这突出了发现参与LSC维持的替代基因靶点和途径的关键需求。药物开发中的网络生物学已成为预测CML疾病药物靶点的重要手段。本计算研究旨在确定BCR-ABL替代分子途径中差异表达的关键调控基因，这些基因可能促进白血病干细胞和祖细胞的根除。CML干细胞和祖细胞与正常细胞的比较分析发现了182个差异表达基因（DEGs）。应用加权基因共表达网络算法（Weighted Gene Co-expression Network Algorithm， WGCNA）确定了一个由73个枢纽基因组成的显著基因模块。蛋白质-蛋白质相互作用和富集分析表明，这些基因参与线粒体翻译延伸、类固醇代谢、胆固醇和脂肪酰基辅酶a的生物合成。此外，我们构建了一个由枢纽基因、cml相关转录因子（TFs）和差异表达microrna （dem）组成的三节点调控网络，突出了三个关键调控因子：ELOVL6、SP1 （TF）和miR-1207-5p。为了探索过表达靶基因ELOVL6的治疗潜力，我们对ELOVL6蛋白结构的植物化学化合物进行了高通量虚拟筛选。随后的分子对接、药代动力学、毒性和分子动力学（MD）模拟显示，两种植物化学物质——含胞苷A和chelidimerine——是CML治疗性生物标志物ELOVL6的潜在抑制剂。

{"title":"Exploring phytochemical inhibitors of fatty acid elongase ELOVL6 for targeted treatment of chronic myeloid leukemia: A comprehensive network-based drug discovery approach","authors":"Alvea Tasneem , Manish Singh , Hridoy R. Bairagya , Ravins Dohare","doi":"10.1016/j.compbiomed.2025.111342","DOIUrl":"10.1016/j.compbiomed.2025.111342","url":null,"abstract":"<div><div>Quiescent leukemic stem cells (LSCs) that persist in the bone marrow microenvironment are responsible for chronic myeloid leukemia (CML) relapses and tyrosine kinase inhibitors (TKIs) resistance. This highlights a critical need to uncover alternative gene targets and pathways involved in LSC maintenance. Network biology in drug development has become essential for predicting drug targets in CML disease. This present computational study aims to identify key regulatory genes that are differentially expressed and involved in molecular pathway alternative to BCR-ABL, which may facilitate the eradication of leukemic stem and progenitor cells. Comparative analysis between CML stem and progenitor cells and their normal counterparts revealed 182 differentially expressed genes (DEGs). Applying Weighted Gene Co-expression Network Algorithm (WGCNA) identified a significant gene module comprising 73 hub genes. Protein-protein interaction and enrichment analyses indicated these genes are involved in mitochondrial translation elongation, steroid metabolism, cholesterol, and fatty acyl-CoA biosynthesis. Furthermore, a three-node regulatory network composed of hub genes, CML-associated transcription factors (TFs), and differentially expressed microRNAs (DEMs) was constructed, highlighting three key regulators: ELOVL6, SP1 (TF), and miR-1207-5p. To explore the therapeutic potential of the overexpressed target gene <em>ELOVL6</em>, we performed high-throughput virtual screening of phytochemical compounds against the ELOVL6 protein structure. Subsequent molecular docking, pharmacokinetics, toxicity, and molecular dynamics (MD) simulations revealed two phytochemicals — withaphysalin A and chelidimerine —as potential inhibitors of the ELOVL6 therapeutic biomarker in CML.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111342"},"PeriodicalIF":6.3,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145799489","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}

引用次数: 0

Drug grouping learning for improving evidence-based treatment recommendations 药物分组学习改善循证治疗建议

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-19 DOI: 10.1016/j.compbiomed.2025.111390

Òscar Raya , Xavier Castells , David Ramírez , Beatriz López

Clinical practice guidelines (CPGs) are essential tools that facilitate the translation of the growing body of scientific evidence into clinical practice by providing clinicians with evidence-based recommendations. The first step of CPG development is the formulation of a clinical question involving an intervention of interest. For some interventions, the quantity and quality of the available scientific evidence can vary. This can significantly impact the treatment recommendations. In this work, we present a method for formulating clinical questions involving pharmacological interventions by considering groups of drugs with shared characteristics. This work focuses on drug grouping based on the treatment outcomes desired by both patient and clinician in addition to pharmacological features. To that end, a new method has been presented to learn distances among drugs that is personalized by considering the preferences of users, and an ensemble clustering method is designed to identify the most suitable grouping for each query. We demonstrate our approach in the context of attention deficit hyperactivity disorder (ADHD). Results demonstrate the feasibility of the approach.

临床实践指南（CPGs）是通过向临床医生提供循证建议，促进将越来越多的科学证据转化为临床实践的重要工具。CPG发展的第一步是制定涉及感兴趣的干预的临床问题。对于某些干预措施，现有科学证据的数量和质量可能各不相同。这可以显著影响治疗建议。在这项工作中，我们提出了一种通过考虑具有共同特征的药物组来制定涉及药理干预的临床问题的方法。这项工作的重点是根据患者和临床医生期望的治疗结果以及药理学特征进行药物分组。为此，提出了一种新的方法，通过考虑用户的个性化偏好来学习药物之间的距离，并设计了一种集成聚类方法来为每个查询识别最合适的分组。我们在注意缺陷多动障碍（ADHD）的背景下证明了我们的方法。结果证明了该方法的可行性。

{"title":"Drug grouping learning for improving evidence-based treatment recommendations","authors":"Òscar Raya , Xavier Castells , David Ramírez , Beatriz López","doi":"10.1016/j.compbiomed.2025.111390","DOIUrl":"10.1016/j.compbiomed.2025.111390","url":null,"abstract":"<div><div>Clinical practice guidelines (CPGs) are essential tools that facilitate the translation of the growing body of scientific evidence into clinical practice by providing clinicians with evidence-based recommendations. The first step of CPG development is the formulation of a clinical question involving an intervention of interest. For some interventions, the quantity and quality of the available scientific evidence can vary. This can significantly impact the treatment recommendations. In this work, we present a method for formulating clinical questions involving pharmacological interventions by considering groups of drugs with shared characteristics. This work focuses on drug grouping based on the treatment outcomes desired by both patient and clinician in addition to pharmacological features. To that end, a new method has been presented to learn distances among drugs that is personalized by considering the preferences of users, and an ensemble clustering method is designed to identify the most suitable grouping for each query. We demonstrate our approach in the context of attention deficit hyperactivity disorder (ADHD). Results demonstrate the feasibility of the approach.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111390"},"PeriodicalIF":6.3,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145799490","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}

引用次数: 0

Meta-learners for few-shot weakly-supervised optic disc and cup segmentation on fundus images 基于元学习器的眼底图像弱监督视盘杯分割

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-17 DOI: 10.1016/j.compbiomed.2025.111384

Pandega Abyan Zumarsyah, Igi Ardiyanto, Hanung Adi Nugroho

Automatic optic disc (OD) and optic cup (OC) segmentation is essential for glaucoma diagnosis, but limited labeled data remain a major challenge. We address this by developing meta-learners for few-shot weakly-supervised segmentation (FWS). The weak supervision is in the form of sparse labels where only a few pixels are labeled. We significantly improve existing meta-learners by introducing Omni meta-training that enhances data utilization and diversifies the number of shots. We also develop efficient versions that reduce computational costs while maintaining strong performance. In addition, we develop sparsification techniques that simulate customizable and representative scribbles, points, regions, and other sparse labels. Comprehensive evaluations are performed on DRISHTI-GS, REFUGE, and RIM-ONE r3 datasets. We find that Omni and efficient versions outperform the original versions, with the best meta-learner being Efficient Omni ProtoSeg (EO-ProtoSeg). It achieves intersection over union (IoU) scores of 88.15% for OD and 71.17% for OC on the REFUGE dataset using just one sparsely labeled image, outperforming few-shot and semi-supervised methods that require more labeled images. EO-ProtoSeg is also comparable to unsupervised domain adaptation methods, yet much lighter with less than two million parameters and requires no retraining. The results highlight the potential of FWS as a lightweight and effective approach in low-label scenarios.

自动视盘（OD）和视杯（OC）分割是青光眼诊断的关键，但有限的标记数据仍然是主要的挑战。我们通过开发用于少量弱监督分割（FWS）的元学习器来解决这个问题。弱监督采用稀疏标签的形式，其中只有少数像素被标记。我们通过引入Omni元训练显著改进了现有的元学习器，提高了数据利用率并使射击次数多样化。我们还开发了高效的版本，在保持强大性能的同时降低了计算成本。此外，我们开发了稀疏化技术来模拟可定制的和代表性的涂鸦、点、区域和其他稀疏标签。对DRISHTI-GS、REFUGE和RIM-ONE r3数据集进行综合评估。我们发现Omni和高效版本优于原始版本，最佳元学习器是高效Omni ProtoSeg （EO-ProtoSeg）。在REFUGE数据集上，仅使用一张稀疏标记的图像，OD的IoU分数为88.15%，OC的IoU分数为71.17%，优于需要更多标记图像的few-shot和半监督方法。EO-ProtoSeg也可与无监督域自适应方法相媲美，但更轻，参数少于200万个，不需要再训练。研究结果突出了FWS在低标签情况下作为一种轻量级和有效方法的潜力。

{"title":"Meta-learners for few-shot weakly-supervised optic disc and cup segmentation on fundus images","authors":"Pandega Abyan Zumarsyah, Igi Ardiyanto, Hanung Adi Nugroho","doi":"10.1016/j.compbiomed.2025.111384","DOIUrl":"10.1016/j.compbiomed.2025.111384","url":null,"abstract":"<div><div>Automatic optic disc (OD) and optic cup (OC) segmentation is essential for glaucoma diagnosis, but limited labeled data remain a major challenge. We address this by developing meta-learners for few-shot weakly-supervised segmentation (FWS). The weak supervision is in the form of sparse labels where only a few pixels are labeled. We significantly improve existing meta-learners by introducing Omni meta-training that enhances data utilization and diversifies the number of shots. We also develop efficient versions that reduce computational costs while maintaining strong performance. In addition, we develop sparsification techniques that simulate customizable and representative scribbles, points, regions, and other sparse labels. Comprehensive evaluations are performed on DRISHTI-GS, REFUGE, and RIM-ONE r3 datasets. We find that Omni and efficient versions outperform the original versions, with the best meta-learner being Efficient Omni ProtoSeg (EO-ProtoSeg). It achieves intersection over union (IoU) scores of 88.15% for OD and 71.17% for OC on the REFUGE dataset using just one sparsely labeled image, outperforming few-shot and semi-supervised methods that require more labeled images. EO-ProtoSeg is also comparable to unsupervised domain adaptation methods, yet much lighter with less than two million parameters and requires no retraining. The results highlight the potential of FWS as a lightweight and effective approach in low-label scenarios.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111384"},"PeriodicalIF":6.3,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145760860","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}

引用次数: 0

A filter-level explainability framework for CNNs in histopathology image analysis 组织病理图像分析中cnn的滤波器级可解释性框架

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-17 DOI: 10.1016/j.compbiomed.2025.111392

Selman Uzun , Gungor Yildirim

Convolutional neural networks (CNNs) have achieved remarkable accuracy in histopathology image classification, yet their decision logic remains largely opaque. Most explainability methods, such as Grad-CAM or SHAP, provide only coarse heatmaps, offering limited insight into the role of individual filters. This study introduces a filter-level explainability framework that quantitatively aligns single-filter activations with Grad-CAM outputs using similarity metrics (Pearson correlation, Dice coefficient, mean squared error), integrates progressive ablation experiments to assess filter contributions, and applies region-based scoring to capture intensity, frequency, and spatial distribution. Evaluations are conducted on the LC25000 dataset, a widely used benchmark comprising 25,000 histopathology image patches from lung and colon tissues (benign and malignant classes). The results demonstrate strong classification performance (Accuracy 0.964, F1-score 0.965, ROC-AUC 0.997) and high filter-to-Grad-CAM agreement (≈0.97 Pearson, ≈0.95 Dice), supported by statistical significance testing. In contrast to prior studies that mainly rely on qualitative visualization, the proposed framework delivers a systematic, statistically validated, and multidimensional approach to filter-level interpretability in CNN-based histopathology analysis. While demonstrated on the LC25000 dataset, the methodological contribution is designed in principle to be dataset-agnostic, as it operates on the final convolutional layer irrespective of dataset choice. By moving beyond qualitative heatmaps, the framework offers a rigorous and reproducible pathway to better understanding CNN decisions, with potential to enhance trust, transparency, and clinical adoption of AI in pathology. In addition, we outline key considerations for translating the framework into routine pathology practice.

卷积神经网络（cnn）在组织病理学图像分类中取得了显著的准确性，但其决策逻辑在很大程度上仍然不透明。大多数可解释性方法，如Grad-CAM或SHAP，只提供粗略的热图，对单个过滤器的作用提供有限的见解。本研究引入了一个滤波器级别的可解释性框架，该框架使用相似性度量（Pearson相关性、Dice系数、均方误差）定量地将单个滤波器的激活与Grad-CAM输出进行比对，整合渐进消融实验来评估滤波器的贡献，并应用基于区域的评分来捕获强度、频率和空间分布。评估是在LC25000数据集上进行的，LC25000数据集是一个广泛使用的基准，包括25000个来自肺和结肠组织的组织病理学图像斑块（良性和恶性分类）。结果显示出较强的分类性能（准确率0.964,F1-score 0.965, ROC-AUC 0.997）和较高的filter-to-Grad-CAM一致性（Pearson≈0.97,Dice≈0.95），并得到统计显著性检验的支持。与之前主要依赖定性可视化的研究相比，该框架提供了一个系统的、统计验证的、多维的方法来实现基于cnn的组织病理学分析的过滤级可解释性。虽然在LC25000数据集上进行了演示，但方法贡献原则上被设计为与数据集无关，因为它在最终的卷积层上操作，而不管数据集选择如何。通过超越定性热图，该框架提供了一个严格且可重复的途径，以更好地理解CNN的决策，有可能增强信任、透明度和人工智能在病理学中的临床应用。此外，我们概述了将该框架转化为常规病理实践的关键考虑因素。

{"title":"A filter-level explainability framework for CNNs in histopathology image analysis","authors":"Selman Uzun , Gungor Yildirim","doi":"10.1016/j.compbiomed.2025.111392","DOIUrl":"10.1016/j.compbiomed.2025.111392","url":null,"abstract":"<div><div>Convolutional neural networks (CNNs) have achieved remarkable accuracy in histopathology image classification, yet their decision logic remains largely opaque. Most explainability methods, such as Grad-CAM or SHAP, provide only coarse heatmaps, offering limited insight into the role of individual filters. This study introduces a filter-level explainability framework that quantitatively aligns single-filter activations with Grad-CAM outputs using similarity metrics (Pearson correlation, Dice coefficient, mean squared error), integrates progressive ablation experiments to assess filter contributions, and applies region-based scoring to capture intensity, frequency, and spatial distribution. Evaluations are conducted on the LC25000 dataset, a widely used benchmark comprising 25,000 histopathology image patches from lung and colon tissues (benign and malignant classes). The results demonstrate strong classification performance (Accuracy 0.964, F1-score 0.965, ROC-AUC 0.997) and high filter-to-Grad-CAM agreement (≈0.97 Pearson, ≈0.95 Dice), supported by statistical significance testing. In contrast to prior studies that mainly rely on qualitative visualization, the proposed framework delivers a systematic, statistically validated, and multidimensional approach to filter-level interpretability in CNN-based histopathology analysis. While demonstrated on the LC25000 dataset, the methodological contribution is designed in principle to be dataset-agnostic, as it operates on the final convolutional layer irrespective of dataset choice. By moving beyond qualitative heatmaps, the framework offers a rigorous and reproducible pathway to better understanding CNN decisions, with potential to enhance trust, transparency, and clinical adoption of AI in pathology. In addition, we outline key considerations for translating the framework into routine pathology practice.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111392"},"PeriodicalIF":6.3,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145760859","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}

引用次数: 0

Regional-aware and sequence-informed multi-decoder network for robust brain glioma segmentation in multi-parametric MRI 基于区域感知和序列信息的多解码器网络在多参数MRI中实现脑胶质瘤的鲁棒分割

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-17 DOI: 10.1016/j.compbiomed.2025.111387

Abbas Mohamed Rezk , Abdulkhalek Al-Fakih , Abdullah Shazly , Vivek Kumar Singh , Yun Hwa Roh , Kanghyun Ryu , Mohammed A. Al-masni

Accurate segmentation of glioblastoma subregions from multi-parametric MRI is essential for diagnosis, surgical planning, and treatment monitoring in neuro-oncology. However, effective delineation of surrounding non-enhancing FLAIR hyperintensity, non-enhancing tumor core, and enhancing tumor remains challenging due to heterogeneous imaging characteristics. Existing deep learning models often fail to fully exploit the clinical specificity of individual MRI sequences. This work introduces a novel deep learning framework featuring (1) a multi-decoder architecture that independently segments key tumor subregions, (2) a sequence-informed guidance strategy that aligns each decoder with MRI sequences best suited to its diagnostic target, and (3) a modified self-attention mechanism for enhanced feature recalibration. These innovations enable precise, region-specific segmentation while preserving anatomical coherence. On the BraTS 2023 dataset, the proposed method achieved an average Dice similarity coefficient (DSC) of 0.9009 and a 95th percentile Hausdorff distance (HD95) of 6.61 mm, surpassing state-of-the-art approaches—particularly for enhancing tumor delineation. Comprehensive ablation studies confirm the contribution of each component. Validation across four external datasets (BraTS 2020, BraTS Africa, MRBrainS18, and BraTS 2024 post-treatment) demonstrates strong generalizability, with DSC gains up to 4.09 % in the most challenging scenarios. By integrating clinical insight with methodological innovation, this framework offers a robust, generalizable solution for glioblastoma segmentation, supporting improved personalized treatment planning and outcome assessment.

从多参数MRI中准确分割胶质母细胞瘤亚区对于神经肿瘤学的诊断、手术计划和治疗监测至关重要。然而，由于异质成像特征，有效描绘周围非增强FLAIR高强度，非增强肿瘤核心和增强肿瘤仍然具有挑战性。现有的深度学习模型往往不能充分利用单个MRI序列的临床特异性。这项工作引入了一种新的深度学习框架，其特点是：(1)独立分割关键肿瘤亚区的多解码器架构，(2)序列信息指导策略，将每个解码器与最适合其诊断目标的MRI序列对齐，以及(3)改进的自注意机制，以增强特征重新校准。这些创新使精确，区域特定的分割，同时保持解剖一致性。在BraTS 2023数据集上，该方法的平均Dice相似系数（DSC）为0.9009，第95百分位Hausdorff距离（HD95）为6.61 mm，超过了最先进的方法，特别是在增强肿瘤描绘方面。综合消融研究证实了每个组成部分的贡献。对四个外部数据集（BraTS 2020、BraTS Africa、MRBrainS18和BraTS 2024处理后）的验证显示出很强的通用性，在最具挑战性的情况下，DSC的增益高达4.09%。通过将临床见解与方法学创新相结合，该框架为胶质母细胞瘤分割提供了一个强大的、可推广的解决方案，支持改进的个性化治疗计划和结果评估。

{"title":"Regional-aware and sequence-informed multi-decoder network for robust brain glioma segmentation in multi-parametric MRI","authors":"Abbas Mohamed Rezk , Abdulkhalek Al-Fakih , Abdullah Shazly , Vivek Kumar Singh , Yun Hwa Roh , Kanghyun Ryu , Mohammed A. Al-masni","doi":"10.1016/j.compbiomed.2025.111387","DOIUrl":"10.1016/j.compbiomed.2025.111387","url":null,"abstract":"<div><div>Accurate segmentation of glioblastoma subregions from multi-parametric MRI is essential for diagnosis, surgical planning, and treatment monitoring in neuro-oncology. However, effective delineation of surrounding non-enhancing FLAIR hyperintensity, non-enhancing tumor core, and enhancing tumor remains challenging due to heterogeneous imaging characteristics. Existing deep learning models often fail to fully exploit the clinical specificity of individual MRI sequences. This work introduces a novel deep learning framework featuring (1) a multi-decoder architecture that independently segments key tumor subregions, (2) a sequence-informed guidance strategy that aligns each decoder with MRI sequences best suited to its diagnostic target, and (3) a modified self-attention mechanism for enhanced feature recalibration. These innovations enable precise, region-specific segmentation while preserving anatomical coherence. On the BraTS 2023 dataset, the proposed method achieved an average Dice similarity coefficient (DSC) of 0.9009 and a 95th percentile Hausdorff distance (HD95) of 6.61 mm, surpassing state-of-the-art approaches—particularly for enhancing tumor delineation. Comprehensive ablation studies confirm the contribution of each component. Validation across four external datasets (BraTS 2020, BraTS Africa, MRBrainS18, and BraTS 2024 post-treatment) demonstrates strong generalizability, with DSC gains up to 4.09 % in the most challenging scenarios. By integrating clinical insight with methodological innovation, this framework offers a robust, generalizable solution for glioblastoma segmentation, supporting improved personalized treatment planning and outcome assessment.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"201 ","pages":"Article 111387"},"PeriodicalIF":6.3,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145760861","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}

引用次数: 0

On the performance of multi-fidelity and reduced-dimensional neural emulators for inference of physiological boundary conditions 多保真度和降维神经仿真器在生理边界条件推理中的性能研究。

IF 6.3 2区医学 Q1 BIOLOGY

Computers in biology and medicine

Pub Date : 2025-12-16 DOI: 10.1016/j.compbiomed.2025.111389

Chloe H. Choi , Andrea Zanoni , Daniele E. Schiavazzi , Alison L. Marsden

Solving inverse problems in cardiovascular modeling is particularly challenging due to the high computational cost of running high-fidelity simulations. In this work, we focus on Bayesian parameter estimation and explore different methods to reduce the computational cost of sampling from the posterior distribution by leveraging low-fidelity approximations. A common approach is to construct a surrogate model for the high-fidelity simulation itself. Another is to build a surrogate for the discrepancy between high- and low-fidelity models. This discrepancy, which is often easier to approximate, is modeled with either a fully connected neural network or a nonlinear dimensionality reduction technique that enables surrogate construction in a lower-dimensional space. A third possible approach is to treat the discrepancy between the high-fidelity and surrogate models as random noise and estimate its distribution using normalizing flows. This allows us to incorporate the approximation error into the Bayesian inverse problem by modifying the likelihood function. We validate five different methods which are variations of the above on analytical test cases by comparing them to posterior distributions derived solely from high-fidelity models, assessing both accuracy and computational cost. Finally, we demonstrate our approaches on two cardiovascular examples of increasing complexity: a lumped-parameter Windkessel model and a patient-specific three-dimensional anatomy.

由于运行高保真仿真的高计算成本，解决心血管建模中的逆问题尤其具有挑战性。在这项工作中，我们将重点放在贝叶斯参数估计上，并探索利用低保真度近似值来减少从后验分布中采样的计算成本的不同方法。一种常见的方法是为高保真仿真本身构建代理模型。另一种方法是为高保真度模型和低保真度模型之间的差异建立一个代理。这种差异通常更容易近似，可以用完全连接的神经网络或非线性降维技术来建模，这种技术可以在较低维空间中进行代理构建。第三种可能的方法是将高保真度和代理模型之间的差异视为随机噪声，并使用归一化流估计其分布。这允许我们通过修改似然函数将近似误差合并到贝叶斯反问题中。我们在分析测试用例上验证了五种不同的方法，这些方法是上述方法的变体，通过将它们与仅由高保真模型推导的后验分布进行比较，评估准确性和计算成本。最后，我们在两个日益复杂的心血管例子上展示了我们的方法：集总参数Windkessel模型和患者特定的三维解剖。

{"title":"On the performance of multi-fidelity and reduced-dimensional neural emulators for inference of physiological boundary conditions","authors":"Chloe H. Choi , Andrea Zanoni , Daniele E. Schiavazzi , Alison L. Marsden","doi":"10.1016/j.compbiomed.2025.111389","DOIUrl":"10.1016/j.compbiomed.2025.111389","url":null,"abstract":"<div><div>Solving inverse problems in cardiovascular modeling is particularly challenging due to the high computational cost of running high-fidelity simulations. In this work, we focus on Bayesian parameter estimation and explore different methods to reduce the computational cost of sampling from the posterior distribution by leveraging low-fidelity approximations. A common approach is to construct a surrogate model for the high-fidelity simulation itself. Another is to build a surrogate for the discrepancy between high- and low-fidelity models. This discrepancy, which is often easier to approximate, is modeled with either a fully connected neural network or a nonlinear dimensionality reduction technique that enables surrogate construction in a lower-dimensional space. A third possible approach is to treat the discrepancy between the high-fidelity and surrogate models as random noise and estimate its distribution using normalizing flows. This allows us to incorporate the approximation error into the Bayesian inverse problem by modifying the likelihood function. We validate five different methods which are variations of the above on analytical test cases by comparing them to posterior distributions derived solely from high-fidelity models, assessing both accuracy and computational cost. Finally, we demonstrate our approaches on two cardiovascular examples of increasing complexity: a lumped-parameter Windkessel model and a patient-specific three-dimensional anatomy.</div></div>","PeriodicalId":10578,"journal":{"name":"Computers in biology and medicine","volume":"200 ","pages":"Article 111389"},"PeriodicalIF":6.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145773813","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}

引用次数: 0