Computer methods and programs in biomedicine最新文献_第5页

TPVNet: A domain-aware graph-based framework for reliable multivariate physiological time series classification in healthcare TPVNet：用于医疗保健中可靠的多变量生理时间序列分类的基于领域感知图的框架。

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-19 DOI: 10.1016/j.cmpb.2025.109214

Xinyue Ren , Yuxuan Xiu , Ting Chen , Zhaosheng Yao , Wai Kin (Victor) Chan

Background and Objective:

Multivariate physiological time series classification is essential for healthcare decision support within the Internet of Medical Things (IoMT). However, existing methods often struggle with high noise, non-stationarity, and privacy concerns inherent in medical signals. This study proposes TPVNet, a novel domain-aware framework, to enhance classification accuracy, stability, and privacy protection in IoMT applications.

Methods:

We propose TPVNet, a domain-aware graph-based framework specifically designed for physiological MTSC tasks. TPVNet integrates two key innovations: a Temporal-enhanced limited Penetrable Visibility Graph (TPVG) for converting raw signals into irreversible graph representations with enriched local and global temporal features. Graph Isomorphism Network (GIN) is used for feature learning followed by a channel-wise voting strategy that aligns with clinical diagnostic workflows to improve decision robustness. Experiments are conducted on seven public physiological datasets, comparing TPVNet against eight state-of-the-art baselines. Performance is evaluated using accuracy, recall, precision, F1-score and standard deviation over 10 replicate tests.

Results:

TPVNet demonstrates robust performance, achieving the highest F1-score on 6 out of 7 datasets. In terms of clinical utility, it significantly outperforms state-of-the-art baselines in data-scarce scenarios; notably, on the Atrial Fibrillation (AF) dataset, it boosts accuracy by 22.0%. Moreover, ablation studies confirm a cumulative accuracy gain of 12.7% over the graph baseline, validating the synergistic effectiveness of the proposed TPVG representation and voting mechanism. Furthermore, TPVNet exhibits superior stability, consistently yielding lower standard deviations compared to deep learning baselines.

Conclusions:

TPVNet provides a privacy-aware, accurate, and stable solution for multivariate physiological time series classification. By integrating domain-inspired graph construction and decision fusion, it offers a clinically aligned framework suitable for real-world IoMT applications, bridging the gap between algorithmic design and healthcare needs.

背景与目的：多变量生理时间序列分类对于医疗物联网（IoMT）中的医疗决策支持至关重要。然而，现有的方法往往与医疗信号固有的高噪声、非平稳性和隐私问题作斗争。本文提出了一种新的领域感知框架TPVNet，以提高IoMT应用中的分类准确性、稳定性和隐私保护。方法：我们提出了TPVNet，这是一个专门为生理MTSC任务设计的基于领域感知图的框架。TPVNet集成了两个关键创新：一个时间增强的有限穿透可见性图（TPVG），用于将原始信号转换为具有丰富的局部和全局时间特征的不可逆图表示。图同构网络（GIN）用于特征学习，然后是与临床诊断工作流程一致的通道智能投票策略，以提高决策鲁棒性。实验在7个公共生理数据集上进行，将TPVNet与8个最先进的基线进行比较。使用准确性、召回率、精密度、f1分数和超过10个重复测试的标准偏差来评估性能。结果：TPVNet表现出稳健的性能，在7个数据集中的6个数据集上获得了最高的f1分。就临床效用而言，在数据稀缺的情况下，它明显优于最先进的基线；值得注意的是，在心房颤动（AF）数据集上，它将准确率提高了22.0%。此外，消融研究证实，与图基线相比，累积精度提高了12.7%，验证了所提出的TPVG表示和投票机制的协同有效性。此外，与深度学习基线相比，TPVNet表现出卓越的稳定性，始终产生更低的标准差。结论：TPVNet为多变量生理时间序列分类提供了一种具有隐私意识、准确、稳定的解决方案。通过集成领域启发的图构建和决策融合，它提供了一个适合现实世界IoMT应用的临床一致的框架，弥合了算法设计和医疗保健需求之间的差距。

{"title":"TPVNet: A domain-aware graph-based framework for reliable multivariate physiological time series classification in healthcare","authors":"Xinyue Ren , Yuxuan Xiu , Ting Chen , Zhaosheng Yao , Wai Kin (Victor) Chan","doi":"10.1016/j.cmpb.2025.109214","DOIUrl":"10.1016/j.cmpb.2025.109214","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Multivariate physiological time series classification is essential for healthcare decision support within the Internet of Medical Things (IoMT). However, existing methods often struggle with high noise, non-stationarity, and privacy concerns inherent in medical signals. This study proposes TPVNet, a novel domain-aware framework, to enhance classification accuracy, stability, and privacy protection in IoMT applications.</div></div><div><h3>Methods:</h3><div>We propose TPVNet, a domain-aware graph-based framework specifically designed for physiological MTSC tasks. TPVNet integrates two key innovations: a Temporal-enhanced limited Penetrable Visibility Graph (TPVG) for converting raw signals into irreversible graph representations with enriched local and global temporal features. Graph Isomorphism Network (GIN) is used for feature learning followed by a channel-wise voting strategy that aligns with clinical diagnostic workflows to improve decision robustness. Experiments are conducted on seven public physiological datasets, comparing TPVNet against eight state-of-the-art baselines. Performance is evaluated using accuracy, recall, precision, F1-score and standard deviation over 10 replicate tests.</div></div><div><h3>Results:</h3><div>TPVNet demonstrates robust performance, achieving the highest F1-score on 6 out of 7 datasets. In terms of clinical utility, it significantly outperforms state-of-the-art baselines in data-scarce scenarios; notably, on the Atrial Fibrillation (AF) dataset, it boosts accuracy by 22.0%. Moreover, ablation studies confirm a cumulative accuracy gain of 12.7% over the graph baseline, validating the synergistic effectiveness of the proposed TPVG representation and voting mechanism. Furthermore, TPVNet exhibits superior stability, consistently yielding lower standard deviations compared to deep learning baselines.</div></div><div><h3>Conclusions:</h3><div>TPVNet provides a privacy-aware, accurate, and stable solution for multivariate physiological time series classification. By integrating domain-inspired graph construction and decision fusion, it offers a clinically aligned framework suitable for real-world IoMT applications, bridging the gap between algorithmic design and healthcare needs.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109214"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145827120","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

Random forests for individual treatment effect estimation with the R package ITERF 随机森林对个体处理效果的估计用R包ITERF。

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-16 DOI: 10.1016/j.cmpb.2025.109187

Sami Tabib, Denis Larocque

Background and Objectives:

Treatment effects often vary across individuals within a population. In contexts such as personalized medicine, it is crucial to accurately estimate treatment effects at the individual level. Random forests are among the most popular, versatile, and efficient statistical learning methods. This article introduces the R package ITERF, designed to estimate individual treatment effects using random forests across various settings. In particular, new methods to estimate the maximum treatment effect are introduced.

Methods and Results:

The ITERF package provides methods for estimating treatment effects in two scenarios: (1) survival outcomes with right-censoring and a binary treatment, and (2) continuous outcomes with a continuous treatment. All methods are based on random forests. A simulation study demonstrates that the proposed methods for estimating the maximum treatment effect perform as expected and show considerable promise. An illustration, using real data, that explores the link between sleep duration and cognitive health in the elderly is given.

Conclusion:

The ITERF package offers a fast and user-friendly tool for estimating treatment effect measures using random forests, making it a valuable resource for researchers and practitioners in personalized treatment evaluation.

背景和目的：治疗效果在人群中因人而异。在个性化医疗等情况下，在个体水平上准确估计治疗效果至关重要。随机森林是最流行、最通用、最有效的统计学习方法之一。本文介绍了R包ITERF，它的设计目的是在各种设置中使用随机森林来估计个别处理的效果。特别介绍了估计最大处理效果的新方法。方法和结果：ITERF包提供了在两种情况下评估治疗效果的方法：(1)右删节和二元治疗的生存结果，以及(2)连续治疗的连续结果。所有的方法都基于随机森林。模拟研究表明，所提出的估计最大处理效果的方法达到了预期的效果，并显示出相当大的前景。本文给出了一个使用真实数据的例子，探讨了老年人睡眠时间与认知健康之间的联系。结论：ITERF包提供了一种快速且用户友好的工具，可以使用随机森林来估计治疗效果度量，为研究人员和从业人员进行个性化治疗评估提供了宝贵的资源。

{"title":"Random forests for individual treatment effect estimation with the R package ITERF","authors":"Sami Tabib, Denis Larocque","doi":"10.1016/j.cmpb.2025.109187","DOIUrl":"10.1016/j.cmpb.2025.109187","url":null,"abstract":"<div><h3>Background and Objectives:</h3><div>Treatment effects often vary across individuals within a population. In contexts such as personalized medicine, it is crucial to accurately estimate treatment effects at the individual level. Random forests are among the most popular, versatile, and efficient statistical learning methods. This article introduces the <span>R</span> package <span>ITERF</span>, designed to estimate individual treatment effects using random forests across various settings. In particular, new methods to estimate the maximum treatment effect are introduced.</div></div><div><h3>Methods and Results:</h3><div>The <span>ITERF</span> package provides methods for estimating treatment effects in two scenarios: (1) survival outcomes with right-censoring and a binary treatment, and (2) continuous outcomes with a continuous treatment. All methods are based on random forests. A simulation study demonstrates that the proposed methods for estimating the maximum treatment effect perform as expected and show considerable promise. An illustration, using real data, that explores the link between sleep duration and cognitive health in the elderly is given.</div></div><div><h3>Conclusion:</h3><div>The <span>ITERF</span> package offers a fast and user-friendly tool for estimating treatment effect measures using random forests, making it a valuable resource for researchers and practitioners in personalized treatment evaluation.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109187"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145827169","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

SegRenal: AI-Driven segmentation of frozen sections in transplant kidney biopsies — A comparative analysis of deep learning models 隔离：移植肾活检中冷冻切片的人工智能驱动分割-深度学习模型的比较分析

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2026-01-02 DOI: 10.1016/j.cmpb.2025.109216

Ibrahim Yilmaz , Heba M. Alazab , Fatih Doganay , Bryan Dangott , Sam Albadri , Aziza Nassar , Fadi Salem , Zeynettin Akkus

Background and Objective:

Frozen section evaluation of donor kidney biopsies is vital for determining transplant suitability, yet remains challenging due to interobserver variability and freezing-related artifacts. While deep learning (DL) has been used for permanent sections, its application to frozen tissue is limited. We developed SegRenal, an artificial intelligence (AI)–based segmentation model for automated identification of glomeruli (non-sclerotic and sclerotic), arteries, and interstitial fibrosis and tubular atrophy (IFTA) in hematoxylin and eosin–stained frozen whole-slide images (WSIs). This study focuses on rigorous model adaptation, dataset development, cross-scanner performance evaluation, and integration into a clinical digital pathology workflow.

Methods:

A total of 183 frozen WSIs were collected from two scanners (GT450 and Grundium) and manually annotated by expert renal pathologists. Three encoder–decoder architectures (UNet, ResNet–UNet, and DenseNet–UNet) were trained on patch-level data to compare binary and multiclass segmentation strategies. The best-performing configuration — pre-trained DenseNet with multiclass output — was further evaluated on 21 unseen WSIs scanned with both platforms. Preprocessing included downsampling, patch extraction, and annotation refinement. Performance was assessed using Dice score, precision, recall, and intraclass correlation coefficient (ICC). Bland–Altman analysis and scanner variability experiments were conducted.

Results:

Pre-trained DenseNet multiclass segmentation yielded the best overall performance: Dice scores of 0.95 (glomeruli), 0.90 (sclerotic glomeruli), 0.80 (arteries), and 0.88 (IFTA). Recall reached 99.8% for glomeruli and 100% for arteries. Performance remained consistent across scanners. In several cases, the model detected structures initially missed by manual annotation, later confirmed by pathologists.

Conclusions:

SegRenal accurately segments key renal compartments in frozen biopsies and demonstrates robust cross-scanner performance. By automating tissue quantification, the model reduces variability and turnaround time, supporting fast and consistent intraoperative kidney transplant assessments.

背景和目的：供体肾活检的冷冻切片评估对于确定移植的适用性至关重要，但由于观察者之间的差异和冷冻相关的伪影，仍然具有挑战性。虽然深度学习（DL）已用于永久性切片，但其在冷冻组织中的应用有限。我们开发了一种基于人工智能（AI）的分割模型，用于自动识别苏木精和伊红染色的冷冻全片图像（wsi）中的肾小球（非硬化和硬化）、动脉、间质纤维化和小管萎缩（IFTA）。本研究的重点是严格的模型适应，数据集开发，跨扫描仪性能评估，并整合到临床数字病理工作流程。方法：从两台扫描仪（GT450和Grundium）上收集183例冷冻wsi，并由肾脏病理学专家手工注释。三种编码器-解码器架构（UNet， ResNet-UNet和DenseNet-UNet）在补丁级数据上进行训练，以比较二进制和多类分割策略。性能最好的配置——具有多类输出的预训练DenseNet——在两个平台扫描的21个未见过的wsi上进行了进一步评估。预处理包括下采样、补丁提取和注释细化。使用Dice评分、准确率、召回率和类内相关系数（ICC）评估性能。进行Bland-Altman分析和扫描仪变异性实验。结果：预先训练的DenseNet多类分割产生了最佳的整体性能：Dice评分为0.95（肾小球），0.90（硬化肾小球），0.80（动脉）和0.88 （IFTA）。肾小球的召回率为99.8%，动脉为100%。各个扫描器的性能保持一致。在一些情况下，模型检测到最初被人工注释遗漏的结构，后来被病理学家证实。结论：在冷冻活检中，SegRenal准确地分割了关键的肾间室，并展示了强大的交叉扫描性能。通过自动化组织量化，该模型减少了可变性和周转时间，支持快速和一致的术中肾移植评估。

{"title":"SegRenal: AI-Driven segmentation of frozen sections in transplant kidney biopsies — A comparative analysis of deep learning models","authors":"Ibrahim Yilmaz , Heba M. Alazab , Fatih Doganay , Bryan Dangott , Sam Albadri , Aziza Nassar , Fadi Salem , Zeynettin Akkus","doi":"10.1016/j.cmpb.2025.109216","DOIUrl":"10.1016/j.cmpb.2025.109216","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Frozen section evaluation of donor kidney biopsies is vital for determining transplant suitability, yet remains challenging due to interobserver variability and freezing-related artifacts. While deep learning (DL) has been used for permanent sections, its application to frozen tissue is limited. We developed <em>SegRenal</em>, an artificial intelligence (AI)–based segmentation model for automated identification of glomeruli (non-sclerotic and sclerotic), arteries, and interstitial fibrosis and tubular atrophy (IFTA) in hematoxylin and eosin–stained frozen whole-slide images (WSIs). This study focuses on rigorous model adaptation, dataset development, cross-scanner performance evaluation, and integration into a clinical digital pathology workflow.</div></div><div><h3>Methods:</h3><div>A total of 183 frozen WSIs were collected from two scanners (GT450 and Grundium) and manually annotated by expert renal pathologists. Three encoder–decoder architectures (UNet, ResNet–UNet, and DenseNet–UNet) were trained on patch-level data to compare binary and multiclass segmentation strategies. The best-performing configuration — pre-trained DenseNet with multiclass output — was further evaluated on 21 unseen WSIs scanned with both platforms. Preprocessing included downsampling, patch extraction, and annotation refinement. Performance was assessed using Dice score, precision, recall, and intraclass correlation coefficient (ICC). Bland–Altman analysis and scanner variability experiments were conducted.</div></div><div><h3>Results:</h3><div>Pre-trained DenseNet multiclass segmentation yielded the best overall performance: Dice scores of 0.95 (glomeruli), 0.90 (sclerotic glomeruli), 0.80 (arteries), and 0.88 (IFTA). Recall reached 99.8% for glomeruli and 100% for arteries. Performance remained consistent across scanners. In several cases, the model detected structures initially missed by manual annotation, later confirmed by pathologists.</div></div><div><h3>Conclusions:</h3><div>SegRenal accurately segments key renal compartments in frozen biopsies and demonstrates robust cross-scanner performance. By automating tissue quantification, the model reduces variability and turnaround time, supporting fast and consistent intraoperative kidney transplant assessments.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109216"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921987","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

Impact of different blood flow modes on hemodynamic environment, thrombosis risk and oxygen transport of oxygenators: A numerical simulation study 不同血流方式对充氧器血流动力学环境、血栓形成风险及氧运输影响的数值模拟研究

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1016/j.cmpb.2026.109236

Xingji Fu , Xiaofang Yang , Feilong Hei , Anqiang Sun , Zengsheng Chen

Background and Objective

Pulsatile blood flow is considered more potential for delivering hemodynamic energy and enhancing microcirculatory perfusion in patients compared to non-pulsatile flow in ECMO. This study aims to systematically evaluate the effects of different blood flow modes on the hemodynamic environment, thrombosis risk, and oxygen transport within oxygenators.

Methods

QUADROX-i Adult Oxygenator was investigated using CFD to simulate its hemodynamic environment under different blood flow modes, comprising one non-pulsatile condition and nine pulsatile conditions with varying frequencies and amplitudes. The stasis (ART) and hypercoagulability (C[FXIa]) were used to assess the thrombosis risk and oxygen transport (PO₂) was also analyzed. The dynamic blood volume (DBV) were calculated to reflect the effective volume within the oxygenator.

Results

Under all blood flow modes, the velocity distribution is more uneven in inlet-side and outlet-side transition regions, and the high value is near the inlet and outlet, and becomes lower away from the inlet and outlet. In gas exchange region, the velocity is low and evenly distributed. The region with the highest ART and C[FXIa] are located at the north corner region close to the outlet. The highest PO₂ evenly appears in the region near the outlet. Under pulsatile conditions, As the flow rate increases, the distribution of velocity, ART and C[FXIa] becomes more uneven, and vice verse. Compared to non-pulsatile condition, the period-averaged ART, C[FXIa] and PO₂ become higher, while the DBV decreases under pulsatile conditions. Amplitude has a more significant effect on all parameters than frequency. Higher amplitude results in the higher period-averaged ART, C[FXIa] and PO₂, alongside a lower DBV.

Conclusions

Uneven flow field mainly occurs in the inlet-side and outlet-side transition region, and the uneven degree increases with the higher flow rate, and vice verse. The highest thrombosis risk locates in the north corner region close to the outlet and the highest oxygen transport occurs in the region close to the outlet. Pulsatile flow can enhance oxygen transport but increase thrombosis risk than non-pulsatile flow. Higher amplitude can increase thrombosis risk but improve oxygen transport in the oxygenator. The frequency variation exhibits minimal influence.

背景与目的与非搏动血流相比，在ECMO中搏动血流被认为更有可能传递血液动力学能量并增强患者的微循环灌注。本研究旨在系统评价不同血流模式对氧合器血流动力学环境、血栓形成风险和氧运输的影响。方法采用CFD模拟不同血流模式下的血流动力学环境，包括1种非脉动状态和9种不同频率和幅值的脉动状态。采用停滞（ART）和高凝性（C[FXIa]）评估血栓形成风险，并分析氧转运（PO2）。计算动态血容量（DBV）以反映氧合器内的有效容量。结果在各血流模式下，入口侧和出口侧过渡区流速分布较为不均匀，在入口和出口附近高值，在远离入口和出口处低值。在气体交换区，流速低且分布均匀。ART和C[FXIa]最高的区域位于靠近出口的北角区域。最高PO2均匀地出现在出口附近区域。在脉动工况下，随着流量的增大，速度、ART和C[FXIa]的分布更加不均匀，反之亦然。与非脉动条件相比，脉动条件下的周期平均ART、C[FXIa]和PO2升高，而DBV降低。振幅对各参数的影响比频率更显著。更高的振幅导致更高的周期平均ART， C[FXIa]和PO2，以及更低的DBV。结论不均匀流场主要发生在进口侧和出口侧过渡区，且不均匀程度随流量的增大而增大，反之亦然。血栓形成风险最高的区域位于靠近出口的北角区域，靠近出口的区域氧转运最高。脉动血流可促进氧转运，但比非脉动血流增加血栓形成风险。较高的振幅可增加血栓形成的风险，但可改善氧合器内的氧运输。频率变化的影响最小。

{"title":"Impact of different blood flow modes on hemodynamic environment, thrombosis risk and oxygen transport of oxygenators: A numerical simulation study","authors":"Xingji Fu , Xiaofang Yang , Feilong Hei , Anqiang Sun , Zengsheng Chen","doi":"10.1016/j.cmpb.2026.109236","DOIUrl":"10.1016/j.cmpb.2026.109236","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Pulsatile blood flow is considered more potential for delivering hemodynamic energy and enhancing microcirculatory perfusion in patients compared to non-pulsatile flow in ECMO. This study aims to systematically evaluate the effects of different blood flow modes on the hemodynamic environment, thrombosis risk, and oxygen transport within oxygenators.</div></div><div><h3>Methods</h3><div>QUADROX-i Adult Oxygenator was investigated using CFD to simulate its hemodynamic environment under different blood flow modes, comprising one non-pulsatile condition and nine pulsatile conditions with varying frequencies and amplitudes. The stasis (ART) and hypercoagulability (C[FXIa]) were used to assess the thrombosis risk and oxygen transport (PO<sub>2</sub>) was also analyzed. The dynamic blood volume (DBV) were calculated to reflect the effective volume within the oxygenator.</div></div><div><h3>Results</h3><div>Under all blood flow modes, the velocity distribution is more uneven in inlet-side and outlet-side transition regions, and the high value is near the inlet and outlet, and becomes lower away from the inlet and outlet. In gas exchange region, the velocity is low and evenly distributed. The region with the highest ART and C[FXIa] are located at the north corner region close to the outlet. The highest PO<sub>2</sub> evenly appears in the region near the outlet. Under pulsatile conditions, As the flow rate increases, the distribution of velocity, ART and C[FXIa] becomes more uneven, and vice verse. Compared to non-pulsatile condition, the period-averaged ART, C[FXIa] and PO<sub>2</sub> become higher, while the DBV decreases under pulsatile conditions. Amplitude has a more significant effect on all parameters than frequency. Higher amplitude results in the higher period-averaged ART, C[FXIa] and PO<sub>2</sub>, alongside a lower DBV.</div></div><div><h3>Conclusions</h3><div>Uneven flow field mainly occurs in the inlet-side and outlet-side transition region, and the uneven degree increases with the higher flow rate, and vice verse. The highest thrombosis risk locates in the north corner region close to the outlet and the highest oxygen transport occurs in the region close to the outlet. Pulsatile flow can enhance oxygen transport but increase thrombosis risk than non-pulsatile flow. Higher amplitude can increase thrombosis risk but improve oxygen transport in the oxygenator. The frequency variation exhibits minimal influence.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109236"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973710","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

Machine learning-assisted prognosis of multiple myeloma side population cells via SRGs and OCLR stemness index 基于SRGs和OCLR干细胞指数的机器学习辅助多发性骨髓瘤侧群细胞的预后。

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-29 DOI: 10.1016/j.cmpb.2025.109211

Xufei Xiang , Ruiyi Yang , Jicong Li , Brian Su , Zefang Xu , Yang An

Background and Objective:

Relapse in Multiple Myeloma, driven by therapy-resistant cancer stem cells, necessitates the development of more specific and accurate prognostic models. Existing stemness indices often lack specificity for the unique biology of Multiple Myeloma. This study aimed to develop, validate, and optimize a novel prognostic gene signature derived from Side Population cells, a well-defined cancer stem cell-enriched subpopulation in Multiple Myeloma.

Methods:

A core stemness gene module was identified from Side Population cell transcriptomes (GSE109651) using Weighted Gene Co-expression Network Analysis, guided by a One-Class Logistic Regression-based stemness index. Stemness-Related Gene scores were computed from this module’s key pathways via single-sample Gene Set Enrichment Analysis. A nonlinear programming algorithm was then employed to create an optimally weighted prognostic model. The model’s performance was validated in independent cohorts (The Cancer Genome Atlas – Multiple Myeloma Research Foundation, GSE24080, GSE57317) using Cox proportional hazards modeling, and its clinical relevance was assessed via drug sensitivity (OncoPredict) and immunotherapy response (Tumor Immune Dysfunction and Exclusion) prediction.

Results:

The resulting Stemness-Related Gene score strongly correlated with the established mRNA stemness index (

r = 0.62

,

p < 1 \times 1 0^{- 82}

). The hsa05222 pathway was identified as the dominant prognostic component (

HR = 12.765

,

p < 0.0001

) and was found to specifically modulate chemoresistance. In contrast, the composite Stemness-Related Gene score better predicted immune evasion potential. The final optimally weighted model, integrating these distinct facets, demonstrated superior prognostic accuracy, consistently outperforming existing benchmarks and simpler models across all validation cohorts.

Conclusions:

This Side Population cell-derived, optimally weighted signature is a robust and multifaceted independent prognostic biomarker for Multiple Myeloma. By distinguishing between chemoresistance and immune evasion profiles, this framework provides a valuable tool to guide personalized, cancer stem cell-targeted therapeutic strategies.

背景和目的：多发性骨髓瘤的复发是由治疗耐药的癌症干细胞驱动的，需要开发更具体和准确的预后模型。对于多发性骨髓瘤独特的生物学特性，现有的干性指标往往缺乏特异性。本研究旨在开发、验证和优化一种新的预后基因特征，该基因来自侧群细胞，这是多发性骨髓瘤中一个定义明确的癌症干细胞富集亚群。方法：采用加权基因共表达网络分析方法，在基于一类Logistic回归的干性指数指导下，从侧边群体细胞转录组（GSE109651）中鉴定出一个核心干性基因模块。通过单样本基因集富集分析，从该模块的关键通路计算stemness相关基因评分。然后采用非线性规划算法建立最优加权预测模型。采用Cox比例风险模型在独立队列（The Cancer Genome Atlas - Multiple Myeloma Research Foundation, GSE24080, GSE57317）中验证了该模型的性能，并通过药物敏感性（OncoPredict）和免疫治疗反应（肿瘤免疫功能障碍和排斥）预测来评估其临床相关性。结果：stemness - related Gene评分与建立的mRNA干性指数呈强相关（r=0.62, p-82）。hsa05222通路被确定为主要的预后成分（HR=12.765）。结论：这种来自侧群细胞的最佳加权信号是多发性骨髓瘤的一个强大的、多方面的独立预后生物标志物。通过区分化疗耐药和免疫逃避特征，该框架为指导个性化的癌症干细胞靶向治疗策略提供了有价值的工具。

{"title":"Machine learning-assisted prognosis of multiple myeloma side population cells via SRGs and OCLR stemness index","authors":"Xufei Xiang , Ruiyi Yang , Jicong Li , Brian Su , Zefang Xu , Yang An","doi":"10.1016/j.cmpb.2025.109211","DOIUrl":"10.1016/j.cmpb.2025.109211","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Relapse in Multiple Myeloma, driven by therapy-resistant cancer stem cells, necessitates the development of more specific and accurate prognostic models. Existing stemness indices often lack specificity for the unique biology of Multiple Myeloma. This study aimed to develop, validate, and optimize a novel prognostic gene signature derived from Side Population cells, a well-defined cancer stem cell-enriched subpopulation in Multiple Myeloma.</div></div><div><h3>Methods:</h3><div>A core stemness gene module was identified from Side Population cell transcriptomes (GSE109651) using Weighted Gene Co-expression Network Analysis, guided by a One-Class Logistic Regression-based stemness index. Stemness-Related Gene scores were computed from this module’s key pathways via single-sample Gene Set Enrichment Analysis. A nonlinear programming algorithm was then employed to create an optimally weighted prognostic model. The model’s performance was validated in independent cohorts (The Cancer Genome Atlas – Multiple Myeloma Research Foundation, GSE24080, GSE57317) using Cox proportional hazards modeling, and its clinical relevance was assessed via drug sensitivity (OncoPredict) and immunotherapy response (Tumor Immune Dysfunction and Exclusion) prediction.</div></div><div><h3>Results:</h3><div>The resulting Stemness-Related Gene score strongly correlated with the established mRNA stemness index (<span><math><mrow><mi>r</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>62</mn></mrow></math></span>, <span><math><mrow><mi>p</mi><mo><</mo><mn>1</mn><mo>×</mo><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mo>−</mo><mn>82</mn></mrow></msup></mrow></math></span>). The <em>hsa05222</em> pathway was identified as the dominant prognostic component (<span><math><mrow><mi>HR</mi><mo>=</mo><mn>12</mn><mo>.</mo><mn>765</mn></mrow></math></span>, <span><math><mrow><mi>p</mi><mo><</mo><mn>0</mn><mo>.</mo><mn>0001</mn></mrow></math></span>) and was found to specifically modulate chemoresistance. In contrast, the composite Stemness-Related Gene score better predicted immune evasion potential. The final optimally weighted model, integrating these distinct facets, demonstrated superior prognostic accuracy, consistently outperforming existing benchmarks and simpler models across all validation cohorts.</div></div><div><h3>Conclusions:</h3><div>This Side Population cell-derived, optimally weighted signature is a robust and multifaceted independent prognostic biomarker for Multiple Myeloma. By distinguishing between chemoresistance and immune evasion profiles, this framework provides a valuable tool to guide personalized, cancer stem cell-targeted therapeutic strategies.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109211"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145877990","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

Online calibration of focal spot drift for a high-resolution micro-CT system 高分辨率微ct系统焦斑漂移的在线标定

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-18 DOI: 10.1016/j.cmpb.2025.109210

Li Chen , Qingxian Zhao , Sinong Su , Yuchen Lu , Yikun Zhang , Shouhua Luo , Yang Chen , Xu Ji

Background:

Micro-CT (Micro-Computed Tomography) is a high-resolution, non-destructive three-dimensional imaging technology widely applied in biomedical research. However, the long scanning time of micro-CT and its higher sensitivity to small-scale perturbations make focal spot drift a more likely and non-negligible source of geometric artifacts. Focal spot drift is typically stochastic and unrepeatable, which may make offline calibration inaccurate, while conventional online calibration tends to be time-consuming due to iterative operations.

Methods:

This paper presents a fast, accurate, and convenient online calibration method that overcomes the limitations commonly associated with existing online calibration approaches. To the best of our knowledge, this work is the first to explicitly and quantitatively describe the relationship between 3D focal spot drift and the resulting 2D projection shifts in micro-CT systems. By leveraging the prior geometric information of a specific feature point on the marker, its true spatial location can be precisely determined, which enables the tracking of its ideal trajectory and corresponding ideal projection positions across all views. Consequently, artifacts induced by focal spot drift can be effectively corrected by compensating for the offsets between the measured and ideal projection positions of the point. The entire correction process requires only a single pass of forward and backward projection.

Results:

The effectiveness and applicability of the method were validated through numerical simulation, physical experiments, and supplementary experiments. In numerical simulations, the method remained effective with even fivefold the normal perturbation level. In physical experiments without ground truth, this method achieved the highest level score according to BRISQUE (Blind/Referenceless Image Spatial Quality Evaluator) and completed the correction in the shortest time, making it the most efficient among all methods. Finally, the supplementary experiments were conducted to verify the applicability of the algorithm under certain assumptions and errors, further demonstrating its practicality.

Conclusions:

The proposed method preserves the high correction accuracy and strong applicability of online calibrations while avoiding the need for frequent iterations or forward/backward projections, thereby achieving high computational efficiency. It is particularly well-suited for micro-CT systems with a large number of scan views and high projection resolution.

背景：微ct （Micro-Computed Tomography, Micro-CT）是一种高分辨率、无损的三维成像技术，广泛应用于生物医学研究。然而，微ct扫描时间长，对小尺度扰动的敏感性高，使得焦点光斑漂移更容易成为几何伪影的一个不可忽略的来源。焦点漂移通常是随机且不可重复的，这可能导致离线校准不准确，而传统的在线校准由于迭代操作而往往耗时。方法：本文提出了一种快速、准确、方便的在线校准方法，克服了现有在线校准方法的局限性。据我们所知，这项工作是第一次明确和定量地描述了微ct系统中三维焦点点漂移和由此产生的二维投影位移之间的关系。通过利用标记上特定特征点的先验几何信息，可以精确地确定其真实空间位置，从而可以在所有视图中跟踪其理想轨迹和相应的理想投影位置。因此，通过补偿点的测量位置和理想投影位置之间的偏移，可以有效地纠正由焦斑漂移引起的伪影。整个校正过程只需要一次前后投影。结果：通过数值模拟、物理实验和补充实验验证了该方法的有效性和适用性。在数值模拟中，即使在正常扰动水平的5倍下，该方法仍然有效。在无ground truth的物理实验中，该方法获得了BRISQUE （Blind/Referenceless Image Spatial Quality Evaluator）的最高评分，并在最短的时间内完成了校正，是所有方法中效率最高的。最后进行了补充实验，验证了算法在一定假设和误差下的适用性，进一步证明了算法的实用性。结论：该方法保留了较高的校正精度和较强的在线标定适用性，避免了频繁的迭代和前向后投影，计算效率较高。它特别适合具有大量扫描视图和高投影分辨率的微型ct系统。

{"title":"Online calibration of focal spot drift for a high-resolution micro-CT system","authors":"Li Chen , Qingxian Zhao , Sinong Su , Yuchen Lu , Yikun Zhang , Shouhua Luo , Yang Chen , Xu Ji","doi":"10.1016/j.cmpb.2025.109210","DOIUrl":"10.1016/j.cmpb.2025.109210","url":null,"abstract":"<div><h3>Background:</h3><div>Micro-CT (Micro-Computed Tomography) is a high-resolution, non-destructive three-dimensional imaging technology widely applied in biomedical research. However, the long scanning time of micro-CT and its higher sensitivity to small-scale perturbations make focal spot drift a more likely and non-negligible source of geometric artifacts. Focal spot drift is typically stochastic and unrepeatable, which may make offline calibration inaccurate, while conventional online calibration tends to be time-consuming due to iterative operations.</div></div><div><h3>Methods:</h3><div>This paper presents a fast, accurate, and convenient online calibration method that overcomes the limitations commonly associated with existing online calibration approaches. To the best of our knowledge, this work is the first to explicitly and quantitatively describe the relationship between 3D focal spot drift and the resulting 2D projection shifts in micro-CT systems. By leveraging the prior geometric information of a specific feature point on the marker, its true spatial location can be precisely determined, which enables the tracking of its ideal trajectory and corresponding ideal projection positions across all views. Consequently, artifacts induced by focal spot drift can be effectively corrected by compensating for the offsets between the measured and ideal projection positions of the point. The entire correction process requires only a single pass of forward and backward projection.</div></div><div><h3>Results:</h3><div>The effectiveness and applicability of the method were validated through numerical simulation, physical experiments, and supplementary experiments. In numerical simulations, the method remained effective with even fivefold the normal perturbation level. In physical experiments without ground truth, this method achieved the highest level score according to BRISQUE (Blind/Referenceless Image Spatial Quality Evaluator) and completed the correction in the shortest time, making it the most efficient among all methods. Finally, the supplementary experiments were conducted to verify the applicability of the algorithm under certain assumptions and errors, further demonstrating its practicality.</div></div><div><h3>Conclusions:</h3><div>The proposed method preserves the high correction accuracy and strong applicability of online calibrations while avoiding the need for frequent iterations or forward/backward projections, thereby achieving high computational efficiency. It is particularly well-suited for micro-CT systems with a large number of scan views and high projection resolution.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109210"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145771863","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

Data-driven bifurcation handling in physics-based reduced-order vascular hemodynamic models 基于物理的降阶血管血流动力学模型中数据驱动的分岔处理

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-30 DOI: 10.1016/j.cmpb.2025.109230

Natalia L. Rubio , Eric F. Darve , Alison L. Marsden

Background and Objective:

Three-dimensional (3D) computational fluid dynamics simulations of cardiovascular flows provide high-fidelity hemodynamic predictions to support cardiovascular medicine, but require substantial computational resources, limiting their clinical applicability. Reduced-order models (ROMs) offer computationally efficient alternatives but suffer from significant accuracy losses, particularly at vessel bifurcations where complex flow physics are inadequately captured by standard Poiseuille flow assumptions. This work presents an enhanced numerical framework that integrates machine learning-predicted bifurcation coefficients into 0D hemodynamic solvers to improve accuracy while maintaining computational efficiency.

Methods:

We develop a resistor–resistor–inductor (RRI) model that uses neural networks to predict pressure-flow relationships from bifurcation geometry, incorporating both linear and quadratic resistance terms along with inductive effects. The method employs physics-based non-dimensionalization to reduce training data requirements and includes flow split prediction for improved geometric characterization. We incorporate the RRI model into a zero-dimensional (0D) cardiovascular flow model using an optimization-based solution strategy. We validate the approach in isolated bifurcations and vascular trees containing up to 40 junctions across Reynolds numbers ranging from 0 to 5500, defining ROM accuracy by comparison to high-fidelity 3D finite element simulation results.

Results:

Results demonstrate substantial accuracy improvements: averaged across all trees and all Reynolds numbers, the RRI method reduces inlet pressure errors from 54 mmHg (45%) for standard 0D models to 25 mmHg (17%), while a simplified resistor-inductor (RI) variant achieves 31 mmHg (26%) error. The enhanced 0D models show particular effectiveness at high Reynolds numbers and in extensive vascular networks.

Conclusions:

This hybrid numerical approach enables accurate, real-time hemodynamic modeling suitable for clinical decision support, uncertainty quantification, and digital twin applications in cardiovascular biomedical engineering.

背景与目的：三维（3D）计算流体动力学模拟心血管血流提供高保真的血流动力学预测，以支持心血管医学，但需要大量的计算资源，限制了其临床适用性。降阶模型（ROMs）提供了计算效率高的替代方案，但存在显著的精度损失，特别是在标准泊泽维尔流假设无法充分捕捉复杂流动物理的船舶分岔处。这项工作提出了一个增强的数值框架，将机器学习预测的分岔系数集成到0D血流动力学求解器中，以提高精度，同时保持计算效率。方法：我们开发了一个电阻-电阻-电感（RRI）模型，该模型使用神经网络从分岔几何中预测压力-流量关系，结合线性和二次电阻项以及感应效应。该方法采用基于物理的无量纲化来减少训练数据的要求，并包括流分裂预测，以改进几何表征。我们使用基于优化的解决方案策略将RRI模型纳入零维（0D）心血管流模型。我们在孤立的分支和血管树中验证了该方法，这些分支和血管树包含多达40个结点，雷诺数范围从0到5500，通过与高保真3D有限元模拟结果进行比较来定义ROM精度。结果表明，在所有树和所有雷诺数的平均值下，RRI方法将进口压力误差从标准0D模型的54 mmHg（45%）减少到25 mmHg(17%)，而简化的电阻-电感器（RI）变体的误差为31 mmHg（26%）。增强的0D模型在高雷诺数和广泛的血管网络中显示出特别的有效性。结论：这种混合数值方法能够实现准确、实时的血流动力学建模，适用于临床决策支持、不确定性量化和心血管生物医学工程中的数字孪生应用。

{"title":"Data-driven bifurcation handling in physics-based reduced-order vascular hemodynamic models","authors":"Natalia L. Rubio , Eric F. Darve , Alison L. Marsden","doi":"10.1016/j.cmpb.2025.109230","DOIUrl":"10.1016/j.cmpb.2025.109230","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Three-dimensional (3D) computational fluid dynamics simulations of cardiovascular flows provide high-fidelity hemodynamic predictions to support cardiovascular medicine, but require substantial computational resources, limiting their clinical applicability. Reduced-order models (ROMs) offer computationally efficient alternatives but suffer from significant accuracy losses, particularly at vessel bifurcations where complex flow physics are inadequately captured by standard Poiseuille flow assumptions. This work presents an enhanced numerical framework that integrates machine learning-predicted bifurcation coefficients into 0D hemodynamic solvers to improve accuracy while maintaining computational efficiency.</div></div><div><h3>Methods:</h3><div>We develop a resistor–resistor–inductor (RRI) model that uses neural networks to predict pressure-flow relationships from bifurcation geometry, incorporating both linear and quadratic resistance terms along with inductive effects. The method employs physics-based non-dimensionalization to reduce training data requirements and includes flow split prediction for improved geometric characterization. We incorporate the RRI model into a zero-dimensional (0D) cardiovascular flow model using an optimization-based solution strategy. We validate the approach in isolated bifurcations and vascular trees containing up to 40 junctions across Reynolds numbers ranging from 0 to 5500, defining ROM accuracy by comparison to high-fidelity 3D finite element simulation results.</div></div><div><h3>Results:</h3><div>Results demonstrate substantial accuracy improvements: averaged across all trees and all Reynolds numbers, the RRI method reduces inlet pressure errors from 54 mmHg (45%) for standard 0D models to 25 mmHg (17%), while a simplified resistor-inductor (RI) variant achieves 31 mmHg (26%) error. The enhanced 0D models show particular effectiveness at high Reynolds numbers and in extensive vascular networks.</div></div><div><h3>Conclusions:</h3><div>This hybrid numerical approach enables accurate, real-time hemodynamic modeling suitable for clinical decision support, uncertainty quantification, and digital twin applications in cardiovascular biomedical engineering.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109230"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145881683","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

MEFA-Unet: Multi-scale feature extraction and fusion attentional unet for segmenting short process of incus in otologic microsurgical scenarios MEFA-Unet：用于耳科显微外科中砧木短过程分割的多尺度特征提取和融合注意网络。

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-02 DOI: 10.1016/j.cmpb.2025.109199

Xin Ding, Yu Huang, Xu Tian, Yang Zhao, Qing Zhang, Zhiqiang Gao, Guodong Feng

Background

The application of convolutional neural networks (CNNs) in microsurgery has been limited. Current CNNs struggle to capture diverse semantic details across various scales and receptive fields, and they also face challenges in establishing meaningful connections between features detected by different receptive fields. Thus, they fail to deal with small objects and maintain accurate boundaries, especially in some complex microsurgical scenarios.

Methods

We propose a multi-scale feature extraction and fusion attentional Unet (MEFA-Unet) to address these limitations. This model employs a hybrid attention mechanism integrating Squeeze-and-Excitation (SE) and Coordinate Attention (CA) modules to enhance feature focus. The MEFA-Unet utilizes multi-scale feature fusion at skip connections and across a multi-layer decoding stage, preserving detailed and contextual information for precise image analysis. For training and validation, the dataset comprising 33,420 human-annotated images from 181 patients was constructed based on four microsurgical scenarios. In addition, 1,500 human-annotated images from 30 raw patients and 2 complex video clips were used to test the networks’ performance.

Results

Compared to other classical segmentation networks, the MEFA-Unet displays superior segmentation performance in visualization results and evaluation metrics. The proposed method achieved a mean intersection over the unit (mIOU) of 0.8880 with the validation set, a mIOU of 0.7583 with the test set, and a mIOU of 0.7855 with the video. Furthermore, the viability of the trained Unet was confirmed by applying it to real microsurgical procedures, where it exhibited excellent performance even in challenging scenarios characterized by weak texture, low contrast, and varying appearances.

Conclusions

The proposed method exhibits a high level of effectiveness in the automated detection and segmentation of the short process of the incus (SPI) across various microsurgical scenarios.

背景：卷积神经网络（cnn）在显微外科中的应用有限。目前的cnn很难捕捉到不同尺度和不同感受野的语义细节，并且在不同感受野检测到的特征之间建立有意义的联系也面临着挑战。因此，它们无法处理小物体并保持准确的边界，特别是在一些复杂的显微手术场景中。方法：我们提出了一种多尺度特征提取和融合注意力Unet （MEFA-Unet）来解决这些局限性。该模型采用挤压激励（SE）和协调注意（CA）模块相结合的混合注意机制来增强特征焦点。MEFA-Unet在跳过连接和多层解码阶段利用多尺度特征融合，为精确的图像分析保留详细的上下文信息。为了训练和验证，基于四种显微外科场景构建了由来自181名患者的33,420张人类注释图像组成的数据集。此外，还使用了来自30名原始患者的1500张人工注释图像和2个复杂视频片段来测试网络的性能。结果：与其他经典分割网络相比，MEFA-Unet在可视化结果和评价指标上表现出优越的分割性能。该方法与验证集的平均单位交点（mIOU）为0.8880，与测试集的mIOU为0.7583，与视频的mIOU为0.7855。此外，经过训练的Unet的可行性通过将其应用于真实的显微外科手术中得到证实，即使在具有弱纹理，低对比度和不同外观的挑战性场景中，它也表现出出色的性能。结论：所提出的方法在各种显微外科场景下对砧骨短过程（SPI）的自动检测和分割中表现出很高的有效性。

{"title":"MEFA-Unet: Multi-scale feature extraction and fusion attentional unet for segmenting short process of incus in otologic microsurgical scenarios","authors":"Xin Ding, Yu Huang, Xu Tian, Yang Zhao, Qing Zhang, Zhiqiang Gao, Guodong Feng","doi":"10.1016/j.cmpb.2025.109199","DOIUrl":"10.1016/j.cmpb.2025.109199","url":null,"abstract":"<div><h3>Background</h3><div>The application of convolutional neural networks (CNNs) in microsurgery has been limited. Current CNNs struggle to capture diverse semantic details across various scales and receptive fields, and they also face challenges in establishing meaningful connections between features detected by different receptive fields. Thus, they fail to deal with small objects and maintain accurate boundaries, especially in some complex microsurgical scenarios.</div></div><div><h3>Methods</h3><div>We propose a multi-scale feature extraction and fusion attentional Unet (MEFA-Unet) to address these limitations. This model employs a hybrid attention mechanism integrating Squeeze-and-Excitation (SE) and Coordinate Attention (CA) modules to enhance feature focus. The MEFA-Unet utilizes multi-scale feature fusion at skip connections and across a multi-layer decoding stage, preserving detailed and contextual information for precise image analysis. For training and validation, the dataset comprising <strong>33,420</strong> human-annotated images from <strong>181</strong> patients was constructed based on four microsurgical scenarios. In addition, <strong>1,500</strong> human-annotated images from <strong>30</strong> raw patients and 2 complex video clips were used to test the networks’ performance.</div></div><div><h3>Results</h3><div>Compared to other classical segmentation networks, the MEFA-Unet displays superior segmentation performance in visualization results and evaluation metrics. The proposed method achieved a mean intersection over the unit (mIOU) of <strong>0.8880</strong> with the validation set, a mIOU of <strong>0.7583</strong> with the test set, and a mIOU of <strong>0.7855</strong> with the video. Furthermore, the viability of the trained Unet was confirmed by applying it to real microsurgical procedures, where it exhibited excellent performance even in challenging scenarios characterized by weak texture, low contrast, and varying appearances.</div></div><div><h3>Conclusions</h3><div>The proposed method exhibits a high level of effectiveness in the automated detection and segmentation of the short process of the incus (SPI) across various microsurgical scenarios.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109199"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145905768","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

Rethinking the value of dynamic and static feature planes in 4D reconstruction of deformable tissues 再思考动、静态特征面在变形组织四维重建中的价值。

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-27 DOI: 10.1016/j.cmpb.2025.109232

Ran Bu , Chenwei Xu , Runyi Liu, Yanzi Miao

Background and Objective:

Reconstructing deformable tissues is crucial for medical image computing and robotic surgery, as it enhances the safety and efficacy of surgical procedures. However, current methods face significant challenges, including errors in tissue reconstruction at occluded regions and limitations in real-time accurate observation of complex structures.

Methods:

In this paper, we present a novel method called Rethink Plane (RPlane), an efficient framework based on Neural Radiance Fields (NeRF), designed to reconstruct global high-fidelity deformable tissues from binocular endoscopic videos efficiently. Our main contribution lies in rethinking the value of dynamic and static features that existing methods often overlook, and developing a Depth Uncertainty Filter. Throughout this work, the dynamic filter is an extremely important foundational component. Based on this, a Dynamic Feature Enhancement module is proposed to address the depth distortion problem caused by the occlusion of surgical instruments. Additionally, a Color Recurrent Refinement strategy is proposed to reduce dynamic blurring caused by instrument contact or tissue self-motion. We validate the effectiveness of RPlane on two datasets (ENDONERF and StereoMIS).

Results:

In all cases, RPlane achieves state-of-the-art (SOTA) performance in terms of tissue reconstruction quality and detail clarity (with a PSNR of 40.527 in ENDONERF and 36.267 in StereoMIS). Furthermore, RPlane demonstrates a 53.3% improvement in robustness without increasing training time or computational resources.

Conclusions:

RPlane addresses depth reconstruction errors caused by surgical instrument occlusion, which are common in existing algorithms. The Dynamic Feature Enhancement module is used to enhance geometric modeling in occluded areas, while the Dynamic Weight Generation & Fusion and Color Recurrent Refinement strategies improve the texture details of tissues. This significant performance improvement promises to be an innovative solution for intraoperative applications.

背景与目的：可变形组织的重建对于医学图像计算和机器人手术至关重要，因为它可以提高手术过程的安全性和有效性。然而，目前的方法面临着巨大的挑战，包括闭塞区域组织重建的误差和复杂结构实时准确观察的局限性。方法：本文提出了一种基于神经辐射场（NeRF）的高效框架——重新思考平面（RPlane），用于从双目内窥镜视频中高效地重建全局高保真的可变形组织。我们的主要贡献在于重新思考现有方法经常忽略的动态和静态特征的价值，并开发了深度不确定性过滤器。在整个工作中，动态滤波器是一个极其重要的基础组件。在此基础上，提出了一种动态特征增强模块来解决手术器械遮挡引起的深度失真问题。此外，提出了一种颜色循环细化策略，以减少由仪器接触或组织自运动引起的动态模糊。我们在两个数据集（ENDONERF和StereoMIS）上验证了RPlane的有效性。结果：在所有病例中，RPlane在组织重建质量和细节清晰度方面达到了最先进的（SOTA）性能（在ENDONERF和StereoMIS中PSNR分别为40.527和36.267）。此外，RPlane在不增加训练时间或计算资源的情况下，鲁棒性提高了53.3%。结论：RPlane解决了现有算法中常见的手术器械闭塞导致的深度重建误差。动态特征增强模块用于增强遮挡区域的几何建模，而动态权值生成与融合和颜色循环细化策略用于改进组织的纹理细节。这种显著的性能改进有望成为术中应用的创新解决方案。

{"title":"Rethinking the value of dynamic and static feature planes in 4D reconstruction of deformable tissues","authors":"Ran Bu , Chenwei Xu , Runyi Liu, Yanzi Miao","doi":"10.1016/j.cmpb.2025.109232","DOIUrl":"10.1016/j.cmpb.2025.109232","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Reconstructing deformable tissues is crucial for medical image computing and robotic surgery, as it enhances the safety and efficacy of surgical procedures. However, current methods face significant challenges, including errors in tissue reconstruction at occluded regions and limitations in real-time accurate observation of complex structures.</div></div><div><h3>Methods:</h3><div>In this paper, we present a novel method called Rethink Plane (RPlane), an efficient framework based on Neural Radiance Fields (NeRF), designed to reconstruct global high-fidelity deformable tissues from binocular endoscopic videos efficiently. Our main contribution lies in rethinking the value of dynamic and static features that existing methods often overlook, and developing a Depth Uncertainty Filter. Throughout this work, the dynamic filter is an extremely important foundational component. Based on this, a Dynamic Feature Enhancement module is proposed to address the depth distortion problem caused by the occlusion of surgical instruments. Additionally, a Color Recurrent Refinement strategy is proposed to reduce dynamic blurring caused by instrument contact or tissue self-motion. We validate the effectiveness of RPlane on two datasets (ENDONERF and StereoMIS).</div></div><div><h3>Results:</h3><div>In all cases, RPlane achieves state-of-the-art (SOTA) performance in terms of tissue reconstruction quality and detail clarity (with a PSNR of 40.527 in ENDONERF and 36.267 in StereoMIS). Furthermore, RPlane demonstrates a 53.3% improvement in robustness without increasing training time or computational resources.</div></div><div><h3>Conclusions:</h3><div>RPlane addresses depth reconstruction errors caused by surgical instrument occlusion, which are common in existing algorithms. The Dynamic Feature Enhancement module is used to enhance geometric modeling in occluded areas, while the Dynamic Weight Generation & Fusion and Color Recurrent Refinement strategies improve the texture details of tissues. This significant performance improvement promises to be an innovative solution for intraoperative applications.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109232"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145910852","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

Application of blockchain-based digital twin technology in healthcare: A scoping review 基于区块链的数字孪生技术在医疗保健中的应用：范围审查

IF 4.8 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2026-03-01 Epub Date: 2025-12-29 DOI: 10.1016/j.cmpb.2025.109231

You Yang, Mengying Liu, Haiying Chen, Li Chen

Background

The integration of blockchain and digital twin (DT) technologies is expected to transform the healthcare sector. DTs are virtual representations of physical entities and allow for real-time monitoring of assets. Predictive analytics of equipment performance can also be supported. Data integrity, security, and trust can be strengthened by blockchain technology. However, the practical applicability and effectiveness of this combined approach in healthcare systems have not been fully established.

Objective

The aim of the present scoping review was to assess the practical applications and synergistic advantages of blockchain-based DT technology in healthcare, evaluate relevant implementation challenges, and provide a research agenda for future studies.

Methods

A scoping review was conducted. PubMed, Web of Science, Scopus, CINAHL, Embase, and OVID were searched systematically. Manual searches were also performed. Boolean operators and targeted keywords were used. Relevant studies were retrieved from database inception to May 20, 2025.

Results

Narrative findings were categorized into three main domains: 1) Technical foundations and core mechanisms for integrating blockchain and DT technologies were described; (2) Application scenarios of blockchain-based DT technology in healthcare were summarized; and (3) Implementation challenges and corresponding solutions for blockchain-based DT technology in healthcare were identified.

Conclusion

The innovative integration of blockchain and DT technologies has advanced the healthcare sector by reshaping the management, interaction, and security of medical data in the digital environment. This convergence establishes a strategic foundation for ongoing digital transformation within healthcare. Future research should prioritize the translation of these developed systems into real-world clinical applications and focus on optimizing their performance to better elucidate how emerging technologies can effectively address practical healthcare challenges.

区块链和数字孪生（DT）技术的集成有望改变医疗保健行业。dt是物理实体的虚拟表示，允许对资产进行实时监控。还可以支持设备性能的预测分析。区块链技术可以增强数据的完整性、安全性和信任度。然而，这种联合方法在医疗保健系统中的实用性和有效性尚未完全建立。本范围综述的目的是评估基于区块链的DT技术在医疗保健中的实际应用和协同优势，评估相关的实施挑战，并为未来的研究提供研究议程。方法进行范围综述。系统检索PubMed、Web of Science、Scopus、CINAHL、Embase和OVID。还执行了手动搜索。使用了布尔运算符和目标关键字。检索自数据库建立至2025年5月20日的相关研究。结果研究结果分为三个主要领域：1)描述了区块链和DT技术集成的技术基础和核心机制；(2)总结了基于区块链的DT技术在医疗领域的应用场景；(3)确定了基于区块链的DT技术在医疗保健领域的实施挑战和相应的解决方案。结论区块链和DT技术的创新整合通过重塑数字环境下医疗数据的管理、交互和安全，推动了医疗行业的发展。这种融合为医疗保健领域正在进行的数字化转型奠定了战略基础。未来的研究应优先考虑将这些已开发的系统转化为现实世界的临床应用，并将重点放在优化其性能上，以更好地阐明新兴技术如何有效地解决实际医疗挑战。

{"title":"Application of blockchain-based digital twin technology in healthcare: A scoping review","authors":"You Yang, Mengying Liu, Haiying Chen, Li Chen","doi":"10.1016/j.cmpb.2025.109231","DOIUrl":"10.1016/j.cmpb.2025.109231","url":null,"abstract":"<div><h3>Background</h3><div>The integration of blockchain and digital twin (DT) technologies is expected to transform the healthcare sector. DTs are virtual representations of physical entities and allow for real-time monitoring of assets. Predictive analytics of equipment performance can also be supported. Data integrity, security, and trust can be strengthened by blockchain technology. However, the practical applicability and effectiveness of this combined approach in healthcare systems have not been fully established.</div></div><div><h3>Objective</h3><div>The aim of the present scoping review was to assess the practical applications and synergistic advantages of blockchain-based DT technology in healthcare, evaluate relevant implementation challenges, and provide a research agenda for future studies.</div></div><div><h3>Methods</h3><div>A scoping review was conducted. PubMed, Web of Science, Scopus, CINAHL, Embase, and OVID were searched systematically. Manual searches were also performed. Boolean operators and targeted keywords were used. Relevant studies were retrieved from database inception to May 20, 2025.</div></div><div><h3>Results</h3><div>Narrative findings were categorized into three main domains: 1) Technical foundations and core mechanisms for integrating blockchain and DT technologies were described; (2) Application scenarios of blockchain-based DT technology in healthcare were summarized; and (3) Implementation challenges and corresponding solutions for blockchain-based DT technology in healthcare were identified.</div></div><div><h3>Conclusion</h3><div>The innovative integration of blockchain and DT technologies has advanced the healthcare sector by reshaping the management, interaction, and security of medical data in the digital environment. This convergence establishes a strategic foundation for ongoing digital transformation within healthcare. Future research should prioritize the translation of these developed systems into real-world clinical applications and focus on optimizing their performance to better elucidate how emerging technologies can effectively address practical healthcare challenges.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"276 ","pages":"Article 109231"},"PeriodicalIF":4.8,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145881241","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