Computer Methods And Programs In Biomedicine[JOURNAL]
Sun
200 papers
A VOF-CFD investigation of mucus transport dynamics with non-Newtonian rheology in a patient-specific tracheobronchial airway across the respiratory cycle.
Tao R, Ding M, Ning J … +3 more , Liu Y, Ma Y, Cai Y
Comput Methods Programs Biomed · 2026 Jun · PMID 42242040 · Publisher ↗
BACKGROUND AND OBJECTIVE: Computational fluid dynamics (CFD) simulations investigating mucus transport in human lung during the resting respiratory cycle remain limited-particularly under pathological conditions such as...
BACKGROUND AND OBJECTIVE: Computational fluid dynamics (CFD) simulations investigating mucus transport in human lung during the resting respiratory cycle remain limited-particularly under pathological conditions such as chronic obstructive pulmonary disease (COPD). METHODS: In this study, a three-dimensional tracheobronchial geometry is reconstructed from computed tomography (CT) images of a patient with COPD. Then, volume of fluid (VOF) model and CFD simulations with distinct pathophysiological scenarios are performed on the airway model to investigate the air-mucus interaction during resting respiration, including the formation of mucus plaque and bronchial muco-obstruction, as well as the dynamics of mucus transport and the efficiency of mucus clearance. The effects of mucus non-Newtonian characteristics on mucus transport are investigated using the Herschel-Bulkley model. In addition, the influences of mucus rheology, ciliary motion, and mucus secretion on airflow and mucus transport are investigated quantitatively. RESULTS: The simulation results demonstrate that there are significant differences (p < 0.05) in the temporal variations of mucus thickness along the airway wall between expiration phase and inspiration phase especially at the trachea bifurcation. The mucus plugging may enhance the mucus viscosity due to the shear-thinning effect and consequently exacerbate the challenges associated with mucus clearance in COPD patients. Due to the pathological changes in mucus production and rheological characteristics, the mucus clearance efficiency in COPD patients is less than half of that in healthy individuals under resting respiratory condition. In addition, the declined ciliary motion exerts detrimental effects on mucus transport and clearance, particularly under condition of high mucus viscosity. CONCLUSIONS: The VOF-CFD model developed in this study enables quantitative analysis of mucus transport dynamics across diverse pathophysiological conditions and can be further refined to support personalized therapeutic assessment through predictions of mucus clearance efficacy.
PyMeshTool - A framework for building efficient automated image-based cardiac anatomical twinning workflows in Python.
Gsell MAF, Klöckl BA, Neic A … +4 more , Mautner B, Augustin CM, Zappon E, Plank G
Comput Methods Programs Biomed · 2026 Jun · PMID 42242039 · Publisher ↗
BACKGROUND AND OBJECTIVE: Digital twin models representing human cardiac electrophysiology have evolved from being primarily research-oriented tools to becoming integral components in clinical decision-making and patient...
BACKGROUND AND OBJECTIVE: Digital twin models representing human cardiac electrophysiology have evolved from being primarily research-oriented tools to becoming integral components in clinical decision-making and patient-specific simulations. The development of such models fundamentally begins with the construction of an accurate anatomical twin, which is typically derived from high-resolution clinical imaging data. This anatomical modeling phase, while essential, is often computationally intensive and time-consuming, necessitating efficient tools to streamline the workflow. To address this need, PyMeshTool, a Python interface for MeshTool, was developed with the primary objective of simplifying and accelerating the anatomical twinning process. METHODS: The C/C++ codebase of MeshTool was restructured to avoid unnecessary source code duplication and to ease the development of the Python interface PyMeshTool. Particular emphasis was placed on the design of PyMeshTool as a streamlined interface that exposes core functionalities of MeshTool in a consistent and user-friendly manner. To evaluate the effectiveness of this design, two Python scripts were implemented running the same anatomical twinning pipeline, one utilizing PyMeshTool and the other calling MeshTool as an external tool. For this purpose, a basic bi-ventricular model generation pipeline was implemented that comprises the generation of simplified universal ventricular coordinates and the assignment of rule-based fibers & sheets. Runtimes and data output of the two workflows were compared for a series of numbers of parallel OpenMP threads. RESULTS: In addition to the primary advantage of PyMeshTool- easy interaction with other Python modules via the Python interface and the compatibility with NumPy- the computational benefits of PyMeshTool have been demonstrated in several comparisons: (i) the PyMeshTool workflow produced less output, 4 files with 89.5MB instead of 418 files with 774.9MB (∼88% less storage usage), (ii) the Python source code was ∼63% shorter in terms of code lines, and (iii) an up to four times faster runtime. CONCLUSION: With the release of the freely available PyMeshTool module, our work aims to streamline image and mesh processing in Python to ease the development of complex pipelines.
Pre-stimulus oculomotor signatures of deception in the Concealed Information Test.
Celniak W, Słapczyńska D, Augustyniak P
Comput Methods Programs Biomed · 2026 May · PMID 42235310 · Publisher ↗
BACKGROUND AND OBJECTIVE: The detection of concealed knowledge related to a crime in forensic settings is usually performed using a traditional polygraph that records physiological responses to specific stimuli. While th...
BACKGROUND AND OBJECTIVE: The detection of concealed knowledge related to a crime in forensic settings is usually performed using a traditional polygraph that records physiological responses to specific stimuli. While these reactive measures are standard, there is growing interest in non-contact behavioral markers that can identify deceptive intent. This study aimed to evaluate the efficacy of anticipatory eye-movement patterns and machine learning classifiers in discriminating between guilty and control participants during the pre-stimulus phase of the examination, identifying cognitive markers that emerge before visual stimuli are even presented. METHODS: To investigate the effectiveness of eye-tracking measurements combined with machine learning in Concealed Information Test, we examined whether anticipatory eye-movement patterns can reliably distinguish between guilty and control participants. At this phase of examination, participants are not yet presented with visual stimuli. We identified features that could reveal subtle differences in cognitive load, arousal, and gaze control associated with the intent to deceive in a series of statistical tests. In our study, several oculomotor features, particularly scanpath-related measures and pupil diameter, were extracted to capture visual exploration patterns and attentional allocation during the anticipation phase of Concealed Information Test. RESULTS: The support vector machine classifier achieved a mean participant-level accuracy of 0.729, which was significantly higher than chance. CONCLUSIONS: These results suggest that anticipatory ocular behavior in CIT may provide valuable insights into cognitive processes associated with anticipation of visual test stimuli even before they are presented.
Transparent augmented-reality head-mounted displays vs medical-grade monitor in laryngologic endoscopy: A comparative FM-100 color discrimination and Fitts' law pointing study.
Dobosz P, Przybyło J
Comput Methods Programs Biomed · 2026 May · PMID 42235309 · Publisher ↗
BACKGROUND: Endoscopic visualization systems are essential in contemporary otolaryngologic surgery. Although high-resolution surgical monitors provide excellent image quality, their size and fixed positioning may comprom...
BACKGROUND: Endoscopic visualization systems are essential in contemporary otolaryngologic surgery. Although high-resolution surgical monitors provide excellent image quality, their size and fixed positioning may compromise ergonomic posture and limit the mobility of endoscopic systems. OBJECTIVE: This study aimed to assess the applicability of transparent augmented-reality Head-Mounted Displays (HMDs) as an alternative or adjunct to conventional monitors, with particular emphasis on color fidelity (FM100) and psychomotor efficiency (Fitts' Law) in simulated endoscopic laryngologic surgery. METHODS: Twenty otolaryngologists experienced in endoscopic sinus surgery evaluated endoscopic images using a standard 40-inch Full HD surgical monitor and two transparent HMDs (Epson Moverio BT-35/BT-40CS). Performance was assessed through a Fitts' Law test and the Farnsworth-Munsell color discrimination test. User feedback on comfort and usability was also collected. Data analysis was conducted using Linear Mixed-Effects Models (LMM), a statistical approach suitable for complex data structures. RESULTS: Statistical analysis revealed a clear speed-accuracy trade-off and a significant performance deficit for HMDs. Both HMDs significantly impaired human color discrimination performance, resulting in a +51% to +60% increase in Total Error Score (TES) compared to the monitor. Furthermore, Fitts' Law modeling demonstrated that while the HMD HD device was faster at low task difficulty (18.9% faster movement time MT baseline), both HMDs were significantly more sensitive to task complexity. The monitor exhibited the most stable performance (MT increased by 6.3% per unit of Index of Difficulty ID), while HMDs showed increased costs (up to 16.4% increase in MT per unit of ID). Participants noted ergonomic improvements but reported limitations in visual alignment and device fit. CONCLUSIONS: Head-mounted displays compromise the fidelity of color perception and lead to diminished efficiency stability under complex surgical task loads, limiting their current applicability for color-critical diagnostic procedures. However, their potential ergonomic advantages support their feasibility in mobile or resource-limited surgical environments where stable color fidelity is not paramount. Further evaluation focusing on improving HMD color accuracy and reducing visual demands is warranted.
PySERA: Open-source standardized python library for automated, scalable, and reproducible handcrafted and deep radiomics.
Salmanpour MR, Pouria AH, Barichin S … +5 more , Salehi Y, Falahati S, Shiri I, Oveisi M, Rahmim A
Comput Methods Programs Biomed · 2026 Sep · PMID 42235274 · Publisher ↗
BACKGROUND: Radiomics analyses extract quantitative biomarkers from medical images for precision modeling, yet reproducibility and scalability remain limited by heterogeneous and limited implementations. Existing tools s...
BACKGROUND: Radiomics analyses extract quantitative biomarkers from medical images for precision modeling, yet reproducibility and scalability remain limited by heterogeneous and limited implementations. Existing tools support only partial standards and lack integration with deep learning (DL) radiomics. To address these gaps, we developed PySERA, an open-source, Python-native, standardized radiomics framework designed for automation, reproducibility, and AI integration. METHODS: PySERA re-implements MATLAB-based SERA (standardized environment for radiomics analysis) in a modular, object-oriented Python architecture. It computes 557 features, including 487 features compliant with the Image Biomarker Standardization Initiative (IBSI) and 10 moment-invariant descriptors, as well as 60 additional diagnostic features, along with DL radiomics embeddings from pre-trained DL: ResNet50 (2048 features) DL radiomics features), DenseNet121 (1024), and VGG16 (512). It includes standardized preprocessing (resampling, discretization, normalization), multi-format I/O (DICOM, NIfTI, NRRD), adaptive memory handling, and a parallel multi-core engine for scalable feature extraction. PySERA integrates directly with libraries: scikit-learn/PyTorch/TensorFlow/MONAI, and others for downstream machine learning applications. RESULTS: PySERA demonstrated >94% IBSI reproducibility, closely matching MITK and substantially outperforming PyRadiomics against the 487 IBSI-compliant feature reference set. Across 8 public datasets, PySERA achieved accuracies of 0.43-0.84, exceeding PyRadiomics for outcome prediction tasks. Benchmarking showed efficient processing (including added higher-order features not implemented in other software): 583 seconds (305 MB) for 166 features, and 2325 seconds (491 MB) for full extraction, with deterministic outputs across platforms. CONCLUSIONS: By uniting standardized handcrafted/DL radiomics in a scalable, transparent, and Python-integrable framework, PySERA establishes a reproducible and extensible foundation for next-generation, AI-ready precision imaging research.
Feasibility of single-beat detection of ventricular late potentials on standard ECG leads via interpretable gradient boosting.
Novak A, Barić D, Zeljković I … +3 more , Penayo B, Pavlović N, Manola Š
Comput Methods Programs Biomed · 2026 May · PMID 42229036 · Publisher ↗
BACKGROUND: Ventricular late potentials (VLPs) are markers of arrhythmogenic substrate, but conventional assessment using signal-averaged ECG (SAECG) requires prolonged acquisition and operator-dependent artifact handlin...
BACKGROUND: Ventricular late potentials (VLPs) are markers of arrhythmogenic substrate, but conventional assessment using signal-averaged ECG (SAECG) requires prolonged acquisition and operator-dependent artifact handling, limiting scalability and ambulatory use. Single-beat detection of VLP-like activity from standard surface ECG remains insufficiently validated. OBJECTIVE: To evaluate the technical feasibility of interpretable single-beat detection of VLP-like perturbations from standard ECG leads without signal averaging. METHODS: Using the MIMIC-IV-ECG database, we analyzed 120,000 beats from leads II, V2, and V6. Because large public datasets with beat-level clinically adjudicated VLP labels are not currently available, physiologically constrained synthetic VLP-like signals were injected into a subset of beats to create a controlled feasibility benchmark. For each beat, more than 200 features were extracted, including time-domain statistics, frequency-domain measures, wavelet coefficients, autocorrelation features, and localized windowed summaries. Ten classifiers were optimized using nested patient-wise cross-validation and evaluated in five settings: single-lead detection, cross-lead generalization, mixed-lead training, reduced training size, and class-imbalance robustness. RESULTS: Gradient-boosted ensembles, particularly XGBoost and CatBoost, achieved strong discrimination on held-out single-beat data (AUC > 0.99; F1 > 0.93), while remaining stable with 10% of the training data and 5% positive-class prevalence. Performance was also robust in lead-transfer experiments. SHAP analysis identified localized entropy, dispersion, and related high-frequency descriptors in late post-R windows as the dominant predictors. CONCLUSION: These findings support the methodological feasibility of interpretable single-beat detection of VLP-like signatures from routine surface ECG under controlled synthetic conditions. Validation on clinically adjudicated cohorts and external datasets is required before clinical translation.
Cell segmentation in microscopy images using a SAM-based U-Net architecture and a novel dataset.
Alam MS, Jackson M, Lord M … +1 more , Meijering E
Comput Methods Programs Biomed · 2026 May · PMID 42224852 · Publisher ↗
BACKGROUND AND OBJECTIVE: Accurate cell instance segmentation is essential for analysing morphology, phenotypes, and dynamics in biomedical research. Challenges such as dense cell populations, blurred boundaries, and ove...
BACKGROUND AND OBJECTIVE: Accurate cell instance segmentation is essential for analysing morphology, phenotypes, and dynamics in biomedical research. Challenges such as dense cell populations, blurred boundaries, and overlapping structures hinder reliable segmentation, and deep learning models remain constrained by the scarcity of fully annotated datasets. METHODS: We present mCellSeg, a dataset of 200 expert-annotated microscopy images containing 16,199 cells from HEK-293T and HUVEC lines, covering diverse morphologies and confluency levels. To utilise this dataset, we propose mSAMUNet, a hybrid neural network that integrates the Segment Anything Model (SAM) with a multiscale U-Net-inspired encoder. This architecture combines the global context modelling of Transformers with the local detail extraction of CNNs, while multiscale branches enhance performance on cells of varying sizes. We benchmarked mSAMUNet against U-Net, StarDist, SAM, and microSAM across four datasets using five-fold cross-validation. RESULTS: mSAMUNet achieved superior performance across all datasets and metrics, including F1 score, SA50, SA75, and mean SA. On the mCellSeg dataset with dense and blurred cell boundaries, mSAMUNet outperformed microSAM (F1 = 0.7071 vs. 0.6994). On the NeurIPS22 dataset, it achieved an F1 score of 0.8675 compared to 0.8483 for microSAM. U-Net and StarDist performed moderately, while SAM was the weakest overall. These results underscore the effectiveness of combining CNN-based multiscale encoders with Transformer-based models for biomedical image segmentation. CONCLUSION: mSAMUNet provides a robust and accurate solution for microscopy cell instance segmentation, surpassing state-of-the-art methods across multiple datasets. Alongside the mCellSeg dataset, this work delivers valuable resources for advancing automated cell analysis with applications in drug discovery, disease modelling, and high-throughput biomedical research.
Memory-augmented graph neural networks for multi-class cardiovascular disease recognition from 12-lead ECG signals.
Li X, Han B, Ma Q … +1 more , Yu Y
Comput Methods Programs Biomed · 2026 May · PMID 42224851 · Publisher ↗
BACKGROUND AND OBJECTIVE: Cardiovascular disease recognition from electrocardiogram (ECG) requires capturing subtle pre-symptomatic electrical abnormalities that manifest across spatially distributed leads and evolve tem...
BACKGROUND AND OBJECTIVE: Cardiovascular disease recognition from electrocardiogram (ECG) requires capturing subtle pre-symptomatic electrical abnormalities that manifest across spatially distributed leads and evolve temporally. Current predictive models inadequately address three critical challenges: the anatomical coupling between ECG leads reflecting cardiac electrical propagation, the multi-scale temporal dynamics spanning millisecond depolarization to circadian rhythms, and the progressive nature of cardiac pathophysiology. We developed an integrated spatiotemporal framework to model these interdependent mechanisms for early cardiovascular risk stratification. METHODS: We implemented a hybrid architecture combining graph convolutions with learnable adjacency matrices to model dynamic inter-lead dependencies based on cardiac electrical topology, hierarchical dilated temporal convolutions capturing rhythms from QRS complexes to heart rate variability patterns, and an external memory bank encoding disease-specific pattern prototypes. Cardiac-aware positional encoding preserved electrophysiological timing relationships, while variational inference quantified recognition uncertainty. We analyzed 10-second 12-lead ECG recordings evaluating myocardial infarction, heart failure, and arrhythmia recognition using area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity with cross-validation. RESULTS: The framework achieved AUROC of 0.94 (sensitivity: 89.2%, specificity: 91.3%) for myocardial infarction, 0.91 (sensitivity: 86.7%, specificity: 88.1%) for heart failure, and 0.96 (sensitivity: 92.8%, specificity: 93.6%) for arrhythmias. Learned graph structures corresponded to known cardiac conduction pathways. Attention mechanisms identified precordial lead interactions during ischemic episodes and temporal windows associated with pathological onset, aligning with clinical prodromal phases. CONCLUSIONS: Explicit modeling of cardiac electrical topology and multi-resolution temporal dynamics enables detection of pre-symptomatic pathophysiological patterns, advancing cardiovascular disease recognition toward proactive risk management with mechanistically interpretable predictions.
Enhancing patient-specific computational modeling of cardiac mechanics by leveraging high spatiotemporal resolution magnetic resonance imaging.
Darayi M, Kara D, Le H … +5 more , Fan Y, Poddar A, Chen D, Nguyen CT, Goswami D
Comput Methods Programs Biomed · 2026 May · PMID 42224850 · Publisher ↗
BACKGROUND AND OBJECTIVE: Computational models of cardiac mechanics can improve diagnosis, surgical planning, and device design; yet their accuracy is limited by the quality of the underlying imaging data used in routine...
BACKGROUND AND OBJECTIVE: Computational models of cardiac mechanics can improve diagnosis, surgical planning, and device design; yet their accuracy is limited by the quality of the underlying imaging data used in routine care. Our team recently reported a free-breathing, four-dimensional cardiac magnetic resonance imaging (MRI) protocol (AutoCMR), which provides 30 volumetric frames per cardiac cycle at 1.6mm isotropic resolution. Here, we investigate whether this high spatiotemporal resolution imaging data, along with cuff systolic and diastolic blood pressures (SBP, DBP), improves non-invasive calibration and internal consistency of patient-specific 0D lumped-parameter and 3D finite element (FE) cardiac models. METHODS: Five adults were imaged with the AutoCMR protocol. Every frame was segmented to obtain left ventricular (LV) and left atrial (LA) volume-time curves. These curves and SBP/DBP calibrated a closed-loop 0D model of LA, LV, and the systemic circulation. To assess the impact on key metrics, the LV elastance up-stroke exponent was varied, while all other calibrated parameters remained fixed. The resulting pressure-volume metrics included end-systolic elastance, effective arterial elastance, ventriculo-arterial coupling, stroke volume, ejection fraction, and stroke work. The calibrated parameters initialized a subject-specific 3D FE LV model. RESULTS: Simulated chamber volumes closely matched MRI, with LV volume root mean square error (RMSE) of 5.55±1.13mL with R=0.951 for 0D; and 7.76±1.65mL with R=0.931 for 3D. Similarly, LA volume RMSE was 3.55±2.61mL with R=0.976 for 0D and 4.32±2.64mL with R=0.973 for 3D. Absolute LV peak-to-cuff systolic pressure target differences averaged 2.81±1.91mmHg (0D) and 4.81±1.10mmHg (3D), while absolute arterial diastolic pressure target differences were 1.64±1.33mmHg (0D) and 1.58±1.51mmHg (3D). In baseline-referenced sensitivity analysis, the 30-frame cine constraints kept mid-systolic LV pressure deviations ∼3-4 mmHg, compared to ∼18-25 mmHg when only end-diastolic and end-systolic constraints were used. CONCLUSIONS: Thirty-frame 3D cine MRI at 1.6mm isotropic resolution, combined with SBP/DBP, supported non-invasive calibration of 0D lumped parameter and 3D FE models with close in-sample agreement to MRI-derived volumes and cuff-pressure targets. The achieved in-sample agreement indicates that this AutoCMR-based simulation approach has the potential to support patient-specific computational assessments of cardiac function in clinical settings.
Stratification of children with myocarditis using radiomics signatures in LGE cardiovascular MRI.
Laube AP, Huellebrand M, Ter-Minassian L … +23 more , Uden T, Schwarzkopf E, Opgen-Rhein B, Anderheiden F, Altmann I, Wiegand G, Boehne M, Reineker K, Rentzsch A, Fischer M, Böcker D, Kaestner M, Hecht T, Schiebel A, Ruf B, Voges I, Gummel K, Pickardt T, Schubert S, Messroghli D, Friede T, Seidel F, Hennemuth A
Comput Methods Programs Biomed · 2026 Sep · PMID 42224791 · Publisher ↗
BACKGROUND AND OBJECTIVE: Pediatric patients with myocarditis present with heterogeneous symptoms, disease courses and outcomes. Late Gadolinium Enhancement (LGE) in cardiovascular magnetic resonance imaging (CMR) is a r...
BACKGROUND AND OBJECTIVE: Pediatric patients with myocarditis present with heterogeneous symptoms, disease courses and outcomes. Late Gadolinium Enhancement (LGE) in cardiovascular magnetic resonance imaging (CMR) is a routine diagnostic tool, but relationships between LGE patterns and patient characteristics are typically assessed qualitatively. We investigate the use of radiomic features to quantify LGE texture and location to identify signatures that stratify pediatric myocarditis cases. METHODS: We compared radiomic features in a digital phantom across different resampling strategies to address variability in patient size and imaging parameters. Non-negative matrix factorization (NMF) was applied to spatially resolved radiomic features of the left myocardium to identify distinct radiomic signatures in a pediatric cohort with confirmed myocarditis. Clinical parameters were compared across the resulting groups, and correlations between image meta-features and outcomes explored. A user-friendly software tool offers feature extraction and signature calculation on unseen data and comparison of new patients to the existing cohort. RESULTS: The phantom experiments showed improved comparability of radiomic features when resampled to uniform voxel density (voxel count per myocardial diameter) rather than uniform voxel size. After appropriate pre-processing, NMF identified four patient groups with distinct LGE signatures within 195 patients (median age 16 years, 19% female). One group separates out patients with signs of heart failure, correlating with left-ventricular ejection fraction (r=-0.38, 95% CI [-0.50,-0.25]) and log(NT-proBNP) (r=0.36,[0.21,0.50]). A second group's dominant meta-feature correlates with myocardial edema (r=0.27,[0.13,0.40]) and ventricular tachycardia (r=0.19,[0.05,0.32]); a third indicates mild presentation. The clinical relevance of the fourth remains unclear. CONCLUSIONS: Spatially resolved radiomic features from suitably resampled LGE CMR images yield quantitative LGE signatures associated with clinical characteristics in pediatric myocarditis, supporting improved stratification and personalized management in the long run.
Animove: a low-cost, easy-to-use, fast and high-resolution system to evaluate swimming performance in zebrafish.
Spineli MN, Vieira PC, Mermelstein C … +1 more , Costa ML
Comput Methods Programs Biomed · 2026 Sep · PMID 42224790 · Publisher ↗
BACKGROUND AND OBJECTIVES: Imaging devices are increasingly present in daily life and are becoming widely used in the record of animal experiments. It is easy to register movement and change in position with video, which...
BACKGROUND AND OBJECTIVES: Imaging devices are increasingly present in daily life and are becoming widely used in the record of animal experiments. It is easy to register movement and change in position with video, which is the basis of several methodologies. Yet most of this data is either only qualitatively analysed or their quantitative analysis is based on expensive, proprietary systems, depending on specialized devices and closed software. Here, we present a novel, low-cost system for quantifying swimming performance in both adult and larval zebrafish (Danio rerio), using smartphones for image acquisition and open source softwares for computational analysis. METHODS: We developed and validated an open-source system for automated, accurate, and sensitive quantification of zebrafish exercise performance, enabling robust behavioral analysis. Adult zebrafish were recorded in a standardized, easy to mount, device using a home-built imaging system, while larval specimens were recorded using transillumination through a standard 6-well plate. We used common smartphones or webcams, since they provide sufficient resolution to capture locomotor dynamics. With the Animove plugin we developed for ImageJ, imported video files were preprocessed using FFmpeg to optimize format compatibility and computational efficiency, enabling analysis on low-specification computer hardware. Animove produces video, images and quantitative data directly, and, associated with Trackmate, can produce quantitative kinematic parameters, including total swimming distance, mean velocity, bout frequency, trajectory patterns, and high-resolution positional tracking. Furthermore, Animove produces several graphical outputs of these results. RESULTS: We used ethanol treatment, whose biphasic effects are well described in both adults and larvae, to validate the system. Ethanol exposure produced stimulant effects in larvae at lower doses (0.5-2%) and depressive effects, leading to sedation, at higher doses (>4%), confirming the system's applicability. CONCLUSIONS: By combining consumer-grade, low-cost hardware and open-source software, the Animove plugin provides an accessible, easy-to-use and adaptable tool that can yield accurate quantitative information. This approach has significant potential to facilitate research in exercise physiology, swimming performance, toxicology, and behavioral science.
DS-MTNet: A dual-stream multi-task network for histopathology localization and histologic risk stratification from small biopsy specimens of bladder cancer.
Fu L, Huang B, Shi L … +5 more , Geng Y, Li J, Cui L, Cao R, Jiang X
Comput Methods Programs Biomed · 2026 Sep · PMID 42224789 · Publisher ↗
Accurate histologic risk stratification of bladder cancer (BC) from whole-slide images (WSIs) remains challenging due to complex cellular spatial organization and the limited capability of conventional convolution-based...
Accurate histologic risk stratification of bladder cancer (BC) from whole-slide images (WSIs) remains challenging due to complex cellular spatial organization and the limited capability of conventional convolution-based methods to capture topological relationships. To address these challenges, this study presents a dual-stream multi-task network, termed DS-MTNet, for knowledge-driven risk stratification and tumor localization in computational pathology. DS-MTNet integrates two complementary feature learning streams: a local stream that captures fine-grained morphological representations from histopathological images, and a generalized stream that models intercellular spatial topology by transforming nuclei into graph-structured representations. The graph stream explicitly encodes cellular interactions using a GraphSAGE-based architecture, enabling effective characterization of the tumor microenvironment. In addition, a multi-task learning framework with a hard weight control mechanism is employed to jointly optimize cancer region detection and risk stratification, thereby reducing task interference and enhancing knowledge sharing across tasks. Extensive experiments conducted on a dataset of 115 BCE WSIs showed that DS-MTNet outperformed the selected comparative methods for histologic risk stratification under the current experimental setting. Ablation studies further confirm the effectiveness of the dual-stream design and the multi-task learning strategy. Moreover, DS-MTNet provides interpretable decision evidence through tumor localization maps, Grad-CAM visualizations, and graph-based explanations, facilitating transparent knowledge extraction from histopathological data. These results indicate that DS-MTNet offers an effective and interpretable knowledge-based framework for histologic risk stratification, highlighting its potential applicability in intelligent decision-support systems for computational pathology.
Ultrasound-guided robotic percutaneous puncture for liver tumors with respiratory gating via attention-enhanced Siamese networks.
Wen T, Zeng Q, Yang Z … +3 more , Wu Y, Du D, Zhou S
Comput Methods Programs Biomed · 2026 Sep · PMID 42214815 · Publisher ↗
BACKGROUND AND OBJECTIVE: Accurate and safe needle insertion is critical in abdominal percutaneous interventions, particularly for liver-targeted procedures. Respiratory-induced organ motion presents a major challenge to...
BACKGROUND AND OBJECTIVE: Accurate and safe needle insertion is critical in abdominal percutaneous interventions, particularly for liver-targeted procedures. Respiratory-induced organ motion presents a major challenge to precision, motivating the need for robotic assistance and real-time motion compensation. METHODS: This study presents a robotic puncture system integrated with ultrasound image-based respiratory gating for improved needle placement. The system incorporates preoperative puncture path planning, intraoperative calibration and image-to-patient registration, robot-assisted needle alignment, and real-time ultrasound monitoring. A channel attention-enhanced fusion Siamese network (SE-FSiamFC) is proposed for robust target tracking within ultrasound image sequences, enabling respiratory phase estimation to guide puncture timing. RESULTS: Phantom experiments demonstrated that the robotic system achieved a calibration time of under 100 s with a calibration accuracy of <1 mm. The average targeting accuracy in phantom-based needle insertions was 1.72 ± 0.25 mm. On the CLUST 2015 ultrasound dataset, the SE-FSiamFC network achieved the best tracking performance on the challenging MED group, with an accuracy of 2.170 ± 1.289 mm. In vivo puncture experiments on two pig models showed that without respiratory gating, the targeting errors were 3.03 ± 0.82 mm and 2.77 ± 0.69 mm, respectively. With respiratory gating, the errors were 1.93 ± 0.33 mm and 1.76 ± 0.29 mm, respectively. CONCLUSIONS: The proposed robotic system enables efficient calibration, accurate tracking, and reliable needle guidance under respiratory motion. The integration of ultrasound-based respiratory gating provides favorable temporal windows for needle insertion, enhancing the safety and precision of abdominal percutaneous interventions.
Automatic detection of uterine contractions before and during labor using EHG: A systematic review.
Acquaviva G, Lensen IS, Peri E … +4 more , van der Hout-van der Jagt MB, van Laar J, Galli A, Mischi M
Comput Methods Programs Biomed · 2026 Sep · PMID 42190447 · Publisher ↗
Uterine activity analysis is valuable for monitoring pregnancy progression and detecting contractions to assess the risk of preterm birth and enable timely intervention in case of adverse events. Traditional monitoring t...
Uterine activity analysis is valuable for monitoring pregnancy progression and detecting contractions to assess the risk of preterm birth and enable timely intervention in case of adverse events. Traditional monitoring tools, such as external palpation, tocography, and intrauterine pressure catheters, are limited by factors like invasiveness, lack of accuracy, or suitability only during labor. Electrohysterography (EHG) is a novel promising, non-invasive alternative; nonetheless, distinguishing contractions from basal uterine activity within EHG signals remains significantly challenging. This systematic review describes and compares automatic uterine contraction detection methods proposed in the literature so far. Following a structured screening process, studies published in English up to March 2026 that address the topic under investigation were considered eligible, while review articles and inaccessible studies were excluded. The search was conducted in accordance with the protocol CRD42025611340 registered in PROSPERO, using PubMed, Scopus, and Web of Science databases. In total, 50 studies were included and evaluated for risk of bias using the QUADAS-2 tool. The employed datasets, pre-processing steps, and data preparation methods were examined, followed by a categorization of all detection techniques into three main groups: thresholding methods, Machine Learning classifiers, and Deep Learning approaches. These were analyzed and qualitatively compared to highlight strengths, limitations, and areas for improvement in EHG-based contraction detection research. Findings indicated that adaptive thresholding methods, especially those based on peak detection and RMS-envelope extraction, achieved high performance values. Machine Learning and Deep Learning approaches also demonstrated strong potential, but required training with large annotated datasets to achieve sufficient robustness and generalizability, which may limit their clinical applicability. However, the interpretation of these results is partly constrained by limitations in the underlying evidence, including the absence of a reliable reference for validating detected contractions, restrictive patient selection criteria, potential bias due to lack of blinding during signal analysis, and substantial heterogeneity across studies precluding direct quantitative comparison.
Knowledge distillation for TBI prognosis: Addressing feature mismatch in heterogeneous clinical datasets.
Wang S, Boutelle MG
Comput Methods Programs Biomed · 2026 Sep · PMID 42190446 · Publisher ↗
BACKGROUND: Traumatic Brain Injury (TBI) remains a major public health concern, requiring accurate prognostic models to support clinical decision-making. While high-performance models have been developed using large, fea...
BACKGROUND: Traumatic Brain Injury (TBI) remains a major public health concern, requiring accurate prognostic models to support clinical decision-making. While high-performance models have been developed using large, feature-rich datasets, their applicability is often limited by the variability and restricted availability of clinical data across hospitals. This study explores knowledge distillation as a strategy to adapt complex models for resource-limited settings with fewer available variables. METHODS: A teacher model was trained on the MIMIC-III dataset using 326 features, while a student model, constrained to 20 features common to both MIMIC-III and eICU, learned from the teacher's probabilistic outputs (soft labels). The student model was designed to approximate the teacher's predictions while enhancing generalization. To address differences in outcome distributions between datasets (MIMIC-III: 20.9% mortality; eICU: 8.4% mortality), isotonic regression calibration was applied to refine predicted probabilities. Model performance was evaluated and compared against baseline approaches. RESULTS: The knowledge distillation model outperformed baseline feature selection approaches, achieving higher accuracy (AUC: 0.864; AUPRC: 0.499), significantly surpassing the feature selection model (AUC: 0.856; AUPRC: 0.458; p = 8.069e-8). Calibration further improved model performance (AUC: 0.872; AUPRC: 0.537; p = 4.413e-15), aligning predicted probabilities with observed outcomes and mitigating dataset distribution disparities. CONCLUSION: Knowledge distillation effectively transfers predictive power from a complex, feature-rich model to a feature-constrained environment, improving TBI prognostication despite substantial feature mismatch. These findings highlight the potential for implementing advanced predictive models in diverse clinical settings, with prospective validation needed to confirm their real-world impact.
Federated learning: A new frontier in the exploration of multi-institutional medical imaging data.
Ciupek D, Malawski M, Pieciak T
Comput Methods Programs Biomed · 2026 Sep · PMID 42190445 · Publisher ↗
Artificial intelligence has transformed the perspective of medical imaging, leading to a genuine technological revolution in modern computer-assisted healthcare systems. However, ubiquitously featured deep learning (DL)...
Artificial intelligence has transformed the perspective of medical imaging, leading to a genuine technological revolution in modern computer-assisted healthcare systems. However, ubiquitously featured deep learning (DL) systems require access to a considerable amount of data, facilitating proper knowledge extraction and generalization. Access to such extensive resources may be hindered due to the time and effort required to convey ethical agreements, set up and carry the acquisition procedures through, and manage the datasets adequately with a particular emphasis on proper anonymization. One of the pivotal challenges in the DL field is data integration from various sources acquired using different hardware vendors, diverse acquisition protocols, experimental setups, and even inter-operator variabilities. In this paper, we review the federated learning (FL) concept that fosters the integration of large-scale heterogeneous datasets from multiple institutions in training DL models. In contrast to a centralized approach, the decentralized FL procedure promotes training DL models while preserving data privacy at each institution involved. We formulate the FL principle and comprehensively review general and specialized medical imaging aggregation and learning algorithms, enabling the generation of a globally generalized model. We meticulously go through the challenges in constructing FL-based systems, such as data and model heterogeneities across the institutions, resilience to potential attacks on data privacy, and the variability in computational and communication resources among the entangled sites that might induce efficiency issues of the entire system. Finally, we explore the up-to-date open frameworks for rapid FL-based algorithm prototyping, comprehensively present real-world implementations of FL systems and shed light on future directions in this intensively growing field.
Patient-specific continuous non-invasive blood pressure waveform in computed tomography derived fractional flow reserve calculation: Comparison with population-based pressure models.
Mirota K, Wicher-Żółtaniecka J, Rzycki M … +3 more , Popiel-Humeniuk I, Kosior A, Ostrowski Z
Comput Methods Programs Biomed · 2026 Sep · PMID 42190444 · Publisher ↗
BACKGROUND AND OBJECTIVE: Fractional flow reserve (FFR) derived from coronary computed tomography angiography (FFR-CT) offers a non-invasive alternative to invasive FFR measurement, the clinical gold standard for assessi...
BACKGROUND AND OBJECTIVE: Fractional flow reserve (FFR) derived from coronary computed tomography angiography (FFR-CT) offers a non-invasive alternative to invasive FFR measurement, the clinical gold standard for assessing the functional significance of coronary stenosis. This study presents and validates the Cardiolens FFR-CT Pro®method that integrates coronary computed tomography angiography (CCTA) images, medical interview data, and continuous non-invasive blood pressure (CNBP) waveform signals to provide patient-specific boundary conditions for computational fluid dynamics (CFD) simulations. The aim of this study is to investigate how pressure morphology affects FFR-CT calculations by comparing the use of patient-specific pressure waveforms versus using systolic and diastolic blood pressure values only, obtained from population based formulas. METHODS: In this study, 132 patients (167 vessels) were included prospectively across 11 centers between 2020 and 2022 and analyzed retrospectively. Cardiolens FFR-CT Pro®using patient-specific pressure waveform boundary conditions, and an FFR-CT population based model. Diagnostic performance metrics (accuracy, specificity, sensitivity) were calculated and compared between methods. RESULTS: The Cardiolens FFR-CT Pro®demonstrated higher accuracy (88.6%), specificity (87.4%) and sensitivity (92.5%) than population based pressure input FFR-CT accuracy (87.3%), specificity (87.3%) and sensitivity (87.5%). CONCLUSIONS: Incorporating patient-specific pressure waveform morphology into boundary conditions improves diagnostic performance compared with population-based formulas alone. These findings underscore the clinical utility of non-invasive FFR-CT, particularly when patient-specific physiological data are available, and highlight the potential for broader adoption of this technique in routine practice.
Noise-Pressure Constrained Liutex method for robust vortex identification in 4D Flow MRI of abdominal aortic aneurysms.
Weng C, He D, Cheng S … +7 more , Yuan D, Wang T, Zeng W, Luo Y, Wang J, Li K, Qiu Y
Comput Methods Programs Biomed · 2026 Sep · PMID 42190443 · Publisher ↗
INTRODUCTION: Accurate identification of vortices is critical for predicting the rupture risk of abdominal aortic aneurysms (AAA). However, existing vortex identification methods perform poorly in noisy 4D Flow MRI data....
INTRODUCTION: Accurate identification of vortices is critical for predicting the rupture risk of abdominal aortic aneurysms (AAA). However, existing vortex identification methods perform poorly in noisy 4D Flow MRI data. This study aims to develop a robust vortex identification framework to address these limitations. METHODS: A new vortex identification method based on Liutex method is proposed that integrates relative pressure information and divergence-based noise estimation to enhance the accuracy and robustness of vortex identification in noisy environments. The performance of the proposed method was validated against conventional techniques (vorticity, Q-criterion, Δ-criterion) and original Liutex using 4D Flow MRI data from 10 AAA patients, with the F score as the evaluation metric. RESULTS: Qualitative visualization and quantitative analysis demonstrated the superior performance of the Noise-Pressure Constrained Liutex (NPC-Liutex) method. It achieved clearer delineation of vortical structures across diverse hemodynamic patterns, with significantly higher F scores (average improvement: 0.058), lower spatial entropy (average improvement: 65.8%) and lower false identification rates (average reduced false identification: 78.6%) compared to existing methods. CONCLUSION: The NPC-Liutex method enables reliable extraction of vortical structures, accurate quantification of vortex intensity, and robust tracking of their dynamic evolution in AAA. By addressing noise sensitivity and shear contamination, this approach offers a clinically viable tool for enhancing hemodynamic risk assessment in AAA using 4D Flow MRI.
Deep learning-driven automatic segmentation of uterine peristalsis in intracavitary electrohysterography signals.
Mira-Tomas JM, Diaz-Martinez A, Alberola-Rubio J … +5 more , Llacer J, Castillón G, Caballero S, Alama P, Garcia-Casado J
Comput Methods Programs Biomed · 2026 Sep · PMID 42190442 · Publisher ↗
BACKGROUND AND OBJECTIVE: Uterine peristalsis, the rhythmic contractions of the inner layer of the myometrium during the menstrual cycle, is more challenging to identify than potent contractions during menstruation, preg...
BACKGROUND AND OBJECTIVE: Uterine peristalsis, the rhythmic contractions of the inner layer of the myometrium during the menstrual cycle, is more challenging to identify than potent contractions during menstruation, pregnancy, or childbirth. Peristalsis plays a significant role in sperm ascension and embryo implantation, making its study valuable for fertility research. Intracavitary Electrohysterography (IC-EHG) is a promising technique for the electrophysiological assessment of uterine activity, but identifying basal, contraction, and artifact segments is a task currently performed by experts consuming substantial time, and is affected by the expert's subjectivity. This study aims to develop a deep learning model to aid clinicians in this segmentation task. METHODS: A total of 306 IC-EHG signals, amounting to a total duration of 9318 min, were collected from three different clinical centers. The model architecture, based on a modified version of the U-Time model, was evaluated using event-oriented evaluation performance metrics (accuracy, recall, precision, and F1-score), episode-oriented evaluation performance metrics (Margin Validation Test, including full detection, partial detection, false detection, non-detection, and others), and typical IC-EHG contraction parameters for signal characterization (root mean square amplitude, contraction frequency, and duration). RESULTS: Event-oriented evaluation performance metrics results indicate accurate classification of basal, contraction, and artifact classes (mean F1-score (%): 94.6, 92.3, and 92.3, respectively). Episode-oriented evaluation performance metric results underscore model's ability to detect consistent events (basal (full + partial detection:) 88.9 + 7.3%, contraction: 84.9 + 7.7%, and artifact: 75.3 + 16.8%). Uterine peristalsis parameters derived from IC-EHG contraction events exhibited low mean absolute errors between manual and model-based segmentations: contraction frequency (4.9%), root mean square (2.8%), and duration (6.8%), along with high agreement between both approaches (ICC ≥ 0.92). CONCLUSIONS: The findings underscore the model's reliability and robustness. The developed deep learning-based model offers significant clinical value by not only saving time in the challenging task of uterine peristalsis segmentation but also reducing expert variability. The developed automatic annotation system has the potential to serve as a valuable instrument in the context of infertility and associated pathologies.
Separating corneal decompensation risk from treatment thresholds in Fuchs' endothelial corneal dystrophy prediction models.
Bair H
Comput Methods Programs Biomed · 2026 Sep · PMID 42184733 · Publisher ↗
Abstract loading — click title to view on PubMed.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe