Computer methods and programs in biomedicine最新文献

{"title":"Real-time tracking and inpainting network with joint learning iterative modules for AR-based DALK surgical navigation","authors":"Weimin Liu ,&nbsp;Junjun Pan ,&nbsp;Liyun Jia ,&nbsp;Sijing Rao ,&nbsp;Jie Zang","doi":"10.1016/j.cmpb.2025.109068","DOIUrl":"10.1016/j.cmpb.2025.109068","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Deep anterior lamellar keratoplasty (DALK) is a widely used treatment for eye diseases and requires accurate and even stitch positions during the suturing process. In this regard, the utilization of Augmented Reality (AR) navigation systems shows promising potential in enhancing the stitching process, and a clear and unoccluded view of the corneal regions would help surgeons better plan the stitching positions.</div></div><div><h3>Methods:</h3><div>In this work, we present a joint-learning and iterative network for AR-based suturing navigation. This network aims to improve the performance of the inpainting under serious occlusion in the suturing process. And it can provide both original instruments and inpainted corneal masks along with inpainted frames. The network is based on feature reuse, iterative modules, and mask propagation structures to greatly reduce the computational cost. For the requirement of end-to-end training, we also propose a novel dataset synthesis method to construct a dataset with both occluded and unoccluded image pairs, along with mask and optical flow annotations. We also develop a novel pipeline based on the grid propagation method and inpainted optical flow outputs to provide clear and stable inpainted frames.</div></div><div><h3>Results:</h3><div>Based on the synthetic datasets, compared to the recent outstanding inpainting networks, our framework reaches a better trade-off between performance and computation efficiency. Our Iter-S model finally gets a mean endpoint error (mEPE) of 1.69, a peak signal-to-noise ratio (PSNR) of 36.86, and a structure similarity index measure (SSIM) of 0.976, along with a low inpainting inference time of 16.26ms. Based on the Iter-S, we construct a novel AR navigation system with a frame rate of around 35.14ms/28FPS on average.</div></div><div><h3>Conclusions:</h3><div>The iterative modules can progressively refine the outputs while providing a favorable trade-off between visual performance and real-time computation efficiency based on the selection of iteration times. Our AR navigation framework can provide stable and accurate tracking outputs with well-inpainted results in real time under severe occlusion conditions, which demonstrates the benefits of guiding the stitching operations of surgeons in corneal surgeries.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109068"},"PeriodicalIF":4.8,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119198","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive gas supply system for membrane oxygenator using online model identification and control in normothermic machine perfusion","authors":"Shiwei Wang ,&nbsp;Junwei Jiang ,&nbsp;Jie Hou ,&nbsp;Xirong Liao ,&nbsp;Zhiyong Huang ,&nbsp;Xiaoyu Li ,&nbsp;Jiang Zhang ,&nbsp;Daidi Zhong ,&nbsp;Pan Yang","doi":"10.1016/j.cmpb.2025.109026","DOIUrl":"10.1016/j.cmpb.2025.109026","url":null,"abstract":"<div><h3>Objective:</h3><div>This work addresses a critical challenge in normothermic machine perfusion (NMP) : the precise and safe control of the oxygenator’s gas supply. The objective is to develop a novel control framework that integrates real-time model identification with adaptive pressure control, aiming to dynamically regulate the partial pressure of oxygen in the blood while preventing critical failure modes like plasma leakage.</div></div><div><h3>Methods:</h3><div>By analyzing the oxygenation process within the artificial lung membrane, we demonstrate that the gas supply system’s input–output behavior can be modeled using a discrete-time autoregressive model with exogenous input (ARX). In this model, the concentrations of both gas and liquid phases are related to the gradients of transmembrane pressure. An online parameter identification employs a forgetting factor recursive least squares (FFRLS) algorithm to control the transmembrane pressure difference. The algorithm enables adaptive tuning of a proportional–integral–derivative (PID) controller, and controller parameters are dynamically updated using real-time model estimates. This adaptive mechanism ensures precise sweep gas pressure regulation. Animal experiments utilizing a prototype extracorporeal membrane oxygenation (ECMO) platform validated the integration of online transmembrane pressure identification and adaptive control.</div></div><div><h3>Result:</h3><div>It achieved rapid setpoint tracking with a settling time of less than 4 s and maintained stable transmembrane pressure with a tracking error of less than ±1 mmHg, even during significant blood pressure fluctuations. Blood gas analysis confirmed the system’s efficacy, successfully modulating PaO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span> to a target normoxic range (90–200 mmHg) while simultaneously preventing plasma leakage, which was observed at excessive pressure differentials.</div></div><div><h3>Conclusion:</h3><div>This study proposed a novel adaptive control framework for NMP oxygenators,demonstrating a strategy that simultaneously ensures oxygenator safety by preventing plasma leakage and enables therapeutic regulation of PaO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span> through on-line model identification, with its clinical potential confirmed in preclinical animal trials. This approach provides a robust foundation for improving organ viability during perfusion and prolonging the functional lifespan of the oxygenator, establishing a new pathway toward safer and more effective organ preservation.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109026"},"PeriodicalIF":4.8,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145079699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automated intravascular ultrasound image processing and quantification of coronary artery anomalies: The AIVUS-CAA software","authors":"Anselm W. Stark ,&nbsp;Pooya Mohammadi Kazaj ,&nbsp;Sebastian Balzer ,&nbsp;Marc Ilic ,&nbsp;Manuel Bergamin ,&nbsp;Ryota Kakizaki ,&nbsp;Andreas Giannopoulos ,&nbsp;Andreas Haeberlin ,&nbsp;Lorenz Räber ,&nbsp;Isaac Shiri ,&nbsp;Christoph Gräni","doi":"10.1016/j.cmpb.2025.109065","DOIUrl":"10.1016/j.cmpb.2025.109065","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Coronary artery anomalies (CAA) with an intramural course are associated with elevated risks of ischemia and sudden cardiac death under stress. Intravascular ultrasound (IVUS) is essential for assessing coronary vessel dynamics in these patients. However, the rarity of such anomalies, along with unique geometric changes in the intramural course and ostium, complicates image analysis, leading to inconsistencies and time-consuming evaluations. Our developed executable, zero/low-code software addresses these limitations by providing automated lumen segmentation and cardiac phase identification in IVUS images acquired during rest and stress protocols.</div></div><div><h3>Methods:</h3><div>The software includes: (1) Automated segmentation of lumen contours trained on 9,418 frames (developed by using human in the loop active learning process) validated on 691 frames and tested on 632 frames, IVUS frames from 76 patients (152 studies) with right CAA using a deep learning (DL) model; (2) Extraction of systolic and diastolic frames via a dual-gating approach combining image- and contour-based methods; and (3) A graphical user interface enabling manual correction of the results. The gating module was validated using a custom flow-loop simulating patient-specific hemodynamics, while segmentation accuracy was assessed via intraclass correlation coefficient (ICC) analysis comparing AI-generated contours with those delineated by experienced readers.</div></div><div><h3>Results:</h3><div>The DL model achieved a mean Dice score of 0.91 (SD: 0.08), sensitivity of 0.95 (SD: 0.12), and specificity of 1.00 (SD: 0.00) on the test set. ICC values for lumen area measurements were 1.00 (95%CI: 1.00–1.00) for rest and 1.00 (95%CI: 1.00–1.00) for stress conditions. The gating module demonstrated excellent reproducibility for identifying systolic and diastolic frames under both conditions (ICC = 1.00 for all).</div></div><div><h3>Conclusions:</h3><div>AIVUS-CAA offers a reliable, automated tool for precise IVUS analysis at rest and during stress, enhancing the evaluation of geometrical changes of coronary vessels in CAA patients and enabling efficient clinical analysis in a streamlined workflow.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109065"},"PeriodicalIF":4.8,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145091394","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessment of the sheetlet thickness in human left ventricular free wall samples using X-ray phase-contrast microtomography","authors":"Zhaorui Li ,&nbsp;Shunli Wang ,&nbsp;Kai Li ,&nbsp;Huazhan Gui ,&nbsp;Feng Yuan ,&nbsp;Patrick Clarysse ,&nbsp;François Varray","doi":"10.1016/j.cmpb.2025.109058","DOIUrl":"10.1016/j.cmpb.2025.109058","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>In the left ventricular (LV) wall, most cardiomyocytes are organized into sheetlets. A good knowledge of the sheetlet arrangement is crucial for understanding ventricular functions.</div></div><div><h3>Methods:</h3><div>In this paper, we introduced a <em>distance-field</em> method to measure the evolution of the thickness of local sheetlets and cleavage planes (CPs) in the laminar structure regions of five human LV free wall transmural samples. The data were acquired using the European Synchrotron Radiation Facility. The high-resolution synchrotron radiation phase-contrast micro-tomography (SR-PCT) imaging with an isotropic spatial resolution of <span><math><mrow><mn>3</mn><mo>.</mo><mn>5</mn><mo>×</mo><mn>3</mn><mo>.</mo><mn>5</mn><mo>×</mo><mn>3</mn><mo>.</mo><mn>5</mn><mi>μ</mi><msup><mrow><mi>m</mi></mrow><mrow><mn>3</mn></mrow></msup></mrow></math></span> allows for a clear observation of the sheetlet arrangement. First, we flattened the samples using Difference of Gaussians (DoG). Secondly, we extracted the sheetlets and CPs using connex filters with different sizes. Then, we generated the laminar structure simulation models to validate <em>distance-field</em> method. Last, we measured the thickness of local CPs and sheetlets by calculating their Chamfer-distance fields.</div></div><div><h3>Results:</h3><div>Sheetlets are thinner and CPs are thicker in the regions with flat sheetlet arrangements; CPs are thicker around blood vessels; and sheetlets are thinner around the intersection region of two sheetlet populations. This regional variation relates to the location of samples in the LV wall, the sheetlet organization manner, and the local myocardial architecture.</div></div><div><h3>Conclusions:</h3><div>The results demonstrate that the distribution of the thickness of sheetlets and CPs is regional, which provides morphology support for the future research about the myocardial mechanical function and the pathological mechanism of heart diseases.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109058"},"PeriodicalIF":4.8,"publicationDate":"2025-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145085333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integration of nested cross-validation, automated hyperparameter optimization, high-performance computing to reduce and quantify the variance of test performance estimation of deep learning models","authors":"Paul Calle ,&nbsp;Averi Bates ,&nbsp;Justin C. Reynolds ,&nbsp;Yunlong Liu ,&nbsp;Haoyang Cui ,&nbsp;Sinaro Ly ,&nbsp;Chen Wang ,&nbsp;Qinghao Zhang ,&nbsp;Alberto J. de Armendi ,&nbsp;Shashank S. Shettar ,&nbsp;Kar-Ming Fung ,&nbsp;Qinggong Tang ,&nbsp;Chongle Pan","doi":"10.1016/j.cmpb.2025.109063","DOIUrl":"10.1016/j.cmpb.2025.109063","url":null,"abstract":"<div><h3>Background and Objectives:</h3><div>The variability and biases in the real-world performance benchmarking of deep learning models for medical imaging compromise their trustworthiness for real-world deployment. The common approach of holding out a single fixed test set fails to quantify the variance in the estimation of test performance metrics. This study introduces NACHOS (Nested and Automated Cross-validation and Hyperparameter Optimization using Supercomputing) to reduce and quantify the variance of test performance metrics of deep learning models.</div></div><div><h3>Methods:</h3><div>NACHOS integrates Nested Cross-Validation (NCV) and Automated Hyperparameter Optimization (AHPO) within a parallelized high-performance computing (HPC) framework. NACHOS was demonstrated on a chest X-ray repository and an Optical Coherence Tomography (OCT) dataset under multiple data partitioning schemes. Beyond performance estimation, DACHOS (Deployment with Automated Cross-validation and Hyperparameter Optimization using Supercomputing) is introduced to leverage AHPO and cross-validation to build the final model on the full dataset, improving expected deployment performance.</div></div><div><h3>Results:</h3><div>The findings underscore the importance of NCV in quantifying and reducing estimation variance, AHPO in optimizing hyperparameters consistently across test folds, and HPC in ensuring computational feasibility.</div></div><div><h3>Conclusions:</h3><div>By integrating these methodologies, NACHOS and DACHOS provide a scalable, reproducible, and trustworthy framework for DL model evaluation and deployment in medical imaging. To maximize public availability, the full open-source codebase is provided at <span><span>https://github.com/thepanlab/NACHOS</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109063"},"PeriodicalIF":4.8,"publicationDate":"2025-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145046609","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep vessel segmentation with U-Net and texture representation of image (TRI) features provides a foundation for improved objective and automated analysis of coronary artery disease from angiography","authors":"Faridoddin Shariaty ,&nbsp;Mobin Mohebi ,&nbsp;Erfan Barzegar-Golmoghani ,&nbsp;Vitalii Pavlov ,&nbsp;Seyed Hamid Aghlmand Sarmi ,&nbsp;Mohammad. H. Behzadpour ,&nbsp;Masume Ahmadi ,&nbsp;Sahar Ramezani Moghadam ,&nbsp;Svetlana Fedyashina ,&nbsp;Ali Mohammadzadeh ,&nbsp;Ali Zahedmehr ,&nbsp;Mohammad Javad Alemzadeh-Ansari ,&nbsp;Ahmad Bitarafan-Rajabi","doi":"10.1016/j.cmpb.2025.109072","DOIUrl":"10.1016/j.cmpb.2025.109072","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Coronary Artery Disease (CAD) diagnosis relies heavily on coronary angiography, yet interpretation suffers from variability. Deep learning (DL) offers potential for improvement, particularly in vessel segmentation, a critical step for analysis. This study aims to enhance vessel segmentation accuracy in angiography using a DL framework incorporating advanced preprocessing and texture features.</div></div><div><h3>Methods</h3><div>We developed a U-Net architecture integrating Texture Representation of Image (TRI) features (Haralick and Law features) to capture subtle vascular details. Advanced preprocessing (Laplacian Pyramid Restoration, Gaussian Differential Scale-Invariance) was applied to improve image quality. The model was pre-trained on the DRIVE dataset and fine-tuned using 7600 clinical angiography images. Performance was evaluated on a held-out test set (19 patients, ∼1700 images) from the same institution and benchmarked against the public ARCADE dataset. Statistical tests assessed performance improvements. Post-segmentation analysis included branching point detection and vessel diameter visualization using heatmaps.</div></div><div><h3>Results</h3><div>The proposed method achieved high segmentation performance on the clinical test set (Accuracy: 0.98, Precision: 0.87, Sensitivity: 0.91, F1-score: 0.89, IoU: 0.801, with CIs provided). Ablation studies confirmed statistically significant contributions from both preprocessing and TRI features (<em>p</em> &lt; 0.01 for all metrics). Performance on the ARCADE benchmark was also strong (F1-score: 0.78), considering annotation differences.</div></div><div><h3>Conclusions</h3><div>Integrating TRI features and advanced preprocessing with a U-Net architecture significantly improves coronary angiography vessel segmentation. This provides a robust foundation for subsequent quantitative analysis potentially supporting CAD assessment. While limitations exist regarding external validation and direct clinical impact assessment, the enhanced segmentation capability represents a valuable advancement for angiographic image analysis tools.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109072"},"PeriodicalIF":4.8,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145106517","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A comprehensive review of techniques, algorithms, advancements, challenges, and clinical applications of multi-modal medical image fusion for improved diagnosis","authors":"Muhammad Zubair ,&nbsp;Muzammil Hussain ,&nbsp;Mousa Ahmad Albashrawi ,&nbsp;Malika Bendechache ,&nbsp;Muhammad Owais","doi":"10.1016/j.cmpb.2025.109014","DOIUrl":"10.1016/j.cmpb.2025.109014","url":null,"abstract":"<div><div>Multi-modal medical image fusion (MMIF) is increasingly recognized as an essential technique for enhancing diagnostic precision and facilitating effective clinical decision-making within computer-aided diagnosis systems. MMIF combines data from X-ray, MRI, CT, PET, SPECT, and ultrasound to create detailed, clinically useful images of patient anatomy and pathology. These integrated representations significantly advance diagnostic accuracy, lesion detection, and segmentation. This comprehensive review meticulously surveys the evolution, methodologies, algorithms, current advancements, and clinical applications of MMIF. We present a critical comparative analysis of traditional fusion approaches, including pixel-, feature-, and decision-level methods, and delves into recent advancements driven by deep learning, generative models, and transformer-based architectures. A critical comparative analysis is presented between these conventional methods and contemporary techniques, highlighting differences in robustness, computational efficiency, and interpretability. The article addresses extensive clinical applications across oncology, neurology, and cardiology, demonstrating MMIF’s vital role in precision medicine through improved patient-specific therapeutic outcomes. Moreover, the review thoroughly investigates the persistent challenges affecting MMIF’s broad adoption, including issues related to data privacy, heterogeneity, computational complexity, interpretability of AI-driven algorithms, and integration within clinical workflows. It also identifies significant future research avenues, such as the integration of explainable AI, adoption of privacy-preserving federated learning frameworks, development of real-time fusion systems, and standardization efforts for regulatory compliance. This review organizes key knowledge, outlines challenges, and highlights opportunities, guiding researchers, clinicians, and developers in advancing MMIF for routine clinical use and promoting personalized healthcare. To support further research, we provide a GitHub repository that includes popular multi-modal medical imaging datasets along with recent models in our shared <span><span>GitHub repository</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109014"},"PeriodicalIF":4.8,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145046511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Asynchronous and focal federated learning for skin lesion classification under local data scarcity and class imbalance","authors":"Shichao Ma ,&nbsp;Yun-Hin Chan ,&nbsp;Edith C.H. Ngai ,&nbsp;Joshua W.K. Ho","doi":"10.1016/j.cmpb.2025.109073","DOIUrl":"10.1016/j.cmpb.2025.109073","url":null,"abstract":"<div><h3>Background and Objectives</h3><div>Federated learning (FL) is an approach that enables the training of machine learning (ML) models using data from multiple data nodes without direct data transfer, hence making it a good choice for healthcare ML applications to alleviate data privacy and security concerns. Most standard FL approaches focus on the setting of a small number of nodes, with each node contributing a sizable amount of data. However, in emerging healthcare settings such as telemedicine and the Internet of Medical Things (IoMT), it is necessary to consider the situation in which there is a large number of nodes, and each contributes a relatively small number (data scarcity) of non-independent (class imbalance) data points.</div></div><div><h3>Methods</h3><div>In this paper, we propose an asynchronous and focal update approach to enable FL to address this problem. In particular, we demonstrate its use in a teledermatology setting, in which a skin lesion image classifier is continuously updated based on data in a highly distributed network of mobile devices. We performed a situation experiment in which 1,268 skin lesion images across 798 mobile devices contributed to the training of a 3-class classifier in an FL framework.</div></div><div><h3>Results</h3><div>We found that widely used synchronous FL methods perform poorly under conditions of data scarcity and imbalance. Specifically, using FedAvg, FedProx, and FedNova, the trained classifiers achieved AUROC values of 0.57-0.67, 0.63-0.66, and 0.64-0.67, respectively, on the held-out test set across various experimental settings. In contrast, our proposed asynchronous and focal approach achieved a test AUROC of 0.78-0.89 after 40 global training epochs. This performance is significantly closer to the optimal AUROC of 0.91, which is achievable by training a classifier with all the data on a centralised server without FL.</div></div><div><h3>Conclusions</h3><div>These results demonstrate that our approach provides a useful solution to implement an efficient FL scheme under the conditions of data scarcity and class imbalance that are commonly found in realistic telemedicine and IoMT applications.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109073"},"PeriodicalIF":4.8,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145046606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Edge computing-based FPGA real-time material decomposition system for photon counting CT","authors":"Mengqing Su ,&nbsp;Xiaopeng Yu ,&nbsp;Qianyu Wu ,&nbsp;Wenhui Qin ,&nbsp;Guotao Quan ,&nbsp;Yanyan Liu ,&nbsp;Wenying Wang ,&nbsp;Yang Chen ,&nbsp;Xiaochun Lai ,&nbsp;Xu Ji","doi":"10.1016/j.cmpb.2025.109040","DOIUrl":"10.1016/j.cmpb.2025.109040","url":null,"abstract":"<div><h3>Background:</h3><div>Photon counting computed tomography (PCCT) has emerged as a potential technology that is revolutionizing clinical CT imaging. Using photon counting detectors (PCDs), the PCCT counts each X-ray event and measures the corresponding energy above the noise floor with significantly higher spatial resolution. However, the multiple-energy-bin setting and much smaller pixels increase the raw data size of PCCT by 20–100 times compared to traditional CT. The current bandwidth of the slip ring struggles to handle such a large volume of data.</div></div><div><h3>Methods:</h3><div>To address the challenge above, we propose an innovative edge computing-based PCCT data processing flow and implement a fast material decomposition program directly on an FPGA system. Our approach eliminates the need to transfer raw data from the CT gantry for later offline processing. Instead, we develop real-time material decomposition and embed the algorithm into an FPGA, which is typically integrated in current PCDs.</div></div><div><h3>Results:</h3><div>Visual and quantitative results on the real phantom dataset have shown that the proposed system can efficiently generate ring-artifact-free and accurate material decomposition results, while also reducing the data volume.</div></div><div><h3>Conclusion:</h3><div>Our proposed system effectively migrates the offline material decomposition processing to the detector blade, thereby fully leveraging the FPGA resources available within the CT gantry to rapidly obtain material decomposition results. Compared to the traditional process flow, our approach offers significantly faster speed and higher throughput, while maintaining the same level of precision as offline processing.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109040"},"PeriodicalIF":4.8,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145027100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CCLR-DL: A novel statistics and deep learning hybrid method for feature selection and forecasting healthcare demand","authors":"Guillem Hernández Guillamet ,&nbsp;Francesc López Seguí ,&nbsp;Josep Vidal Alaball ,&nbsp;Beatriz López","doi":"10.1016/j.cmpb.2025.109057","DOIUrl":"10.1016/j.cmpb.2025.109057","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Hybrid forecasting methods aim to overcome the limitations of classical statistical approaches and deep learning models. While statistical methods provide interpretability, they often lack predictive power. Conversely, deep learning models achieve high accuracy but act as “black boxes.” This study introduces the Comprehensive Cross-Correlation and Lagged Linear Regression Deep Learning (CCLR-DL) framework, combining statistical and deep learning techniques to enhance both forecasting accuracy and interpretability. Unlike existing hybrid methods that combine statistical filtering with deep learning, CCLR-DL integrates causal statistical selection with neural forecasting, producing interpretable predictors and consistently achieving higher accuracy than models without feature selection or other standard baselines.</div></div><div><h3>Methods:</h3><div>The CCLR-DL framework integrates cross-correlation analysis, lagged multiple linear regression, and Granger causality testing with advanced deep learning architectures. This dual-phase approach first identifies causally significant predictors and then fits them into a deep learning model for multivariate time series forecasting. The framework was validated using a real-world dataset of clinical visits and diagnoses from 6.3 million individuals collected over 10 years.</div></div><div><h3>Results:</h3><div>In the evaluated setting, the CCLR-DL framework outperformed baseline models, achieving an average Root Mean Square Error (RMSE) improvement of 19.8% over univariate models, 60.1% over no feature selection, and 51.9% over random selection. The causality phase ensured that all selected predictors demonstrated a significant Granger-causal (GC) relationship. Simpler recurrent architectures, particularly bidirectional Long Short-Term Memory units (BiLSTM), yielded the most accurate forecasts by effectively capturing nonlinear temporal dependencies.</div></div><div><h3>Conclusions:</h3><div>By addressing the challenges of both prediction accuracy and model transparency, the CCLR-DL framework offers a new approach for high-dimensional, multivariate time series forecasting. In healthcare settings, it may enable decision-makers to anticipate demand shifts with greater reliability, allowing earlier staff scheduling, more efficient resource allocation, and reduced waiting times. In our evaluation, it consistently outperformed baseline strategies, delivering measurable improvements that translate into thousands of patient visits being forecasted more accurately across large populations.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"272 ","pages":"Article 109057"},"PeriodicalIF":4.8,"publicationDate":"2025-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145019268","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}