Computer Methods And Programs In Biomedicine[JOURNAL]
Sun
200 papers
Patient-ventilator interaction-Development of a mathematical model of the respiratory center.
Hennigs C, Bilda F, Selpien H … +5 more , Lerg T, Männel G, Becher T, Schädler D, Rostalski P
Comput Methods Programs Biomed · 2026 Jun · PMID 41905158 · Publisher ↗
BACKGROUND AND OBJECTIVE: Patient-ventilator interaction critically influences the safety and effectiveness of mechanical ventilation, particularly under support modes such as pressure support ventilation (PSV). METHODS:...
BACKGROUND AND OBJECTIVE: Patient-ventilator interaction critically influences the safety and effectiveness of mechanical ventilation, particularly under support modes such as pressure support ventilation (PSV). METHODS: This study introduces a mathematical model of the respiratory center that integrates chemical feedback, mechanical feedback, and reflex mechanisms to simulate spontaneous breathing and its interaction with a mechanical ventilator. The model incorporates O and CO chemoreflexes, mechanical feedback, and neural control circuits. RESULTS: Simulation studies evaluated the model's performance under hypercapnia, hypoxia, varying PSV levels, and different types of patient-ventilator asynchrony, comparing outcomes to published literature data. The model accurately reproduced respiratory rate and tidal volume responses with mean absolute percentage errors below 12%, and generated asynchrony frequencies within a 5% acceptance margin. Characteristic ventilatory patterns and emergent asynchronies appeared naturally, without explicit programming of events, demonstrating the physiological plausibility of the model. CONCLUSIONS: This respiratory center model provides a robust tool for investigating respiratory control, testing ventilator strategies, and supporting the development of intelligent ventilation approaches, including in silico clinical trials. Its ability to reproduce both normal and pathological ventilatory patterns underscores its potential for translational and clinical research applications.
Discrimination of vascular proliferation in eyelid surgery using multimodal hyperspectral imaging technology.
Tian C, Jiang Z, Gu X … +5 more , Lv Q, Liu X, Deng J, An G, Li W
Comput Methods Programs Biomed · 2026 Jun · PMID 41905157 · Publisher ↗
BACKGROUND AND OBJECTIVE: Vascular proliferation is a common skin reaction following eyelid surgery, and early diagnosis is crucial for improving treatment outcomes. This study proposes a multimodal framework for the dia...
BACKGROUND AND OBJECTIVE: Vascular proliferation is a common skin reaction following eyelid surgery, and early diagnosis is crucial for improving treatment outcomes. This study proposes a multimodal framework for the diagnosis of vascular proliferation in eyelid tissues, in which hyperspectral images provide spectral-spatial information related to microvascular changes, while corresponding pathological slice images provide complementary microscopic morphological information. These two modalities were jointly analyzed using the HyperT-net deep learning model. METHODS: Hyperspectral data ranging from 400 to 1000 nm and corresponding pathological slice images were collected for model development and evaluation. A combination of one-dimensional convolutional networks and Transformer architecture was employed to build the HyperT-net model for data analysis. RESULTS: The proposed approach effectively captures subtle changes in skin microvasculature, providing a non-invasive and precise tool for skin condition assessment. Compared to existing methods, our approach demonstrates superior accuracy and sensitivity, particularly in the early detection of vascular proliferation. Additionally, ablation experiments reveal that each component of the model plays a critical role in enhancing its performance. CONCLUSIONS: This research introduces a novel technological pathway for diagnosing vascular proliferation in the field of medical aesthetics and lays the foundation for the future promotion of related applications.
Microbubble cavitation in low-intensity focused ultrasound stimulation: A multi-scale computational study.
Yu JH, Han M, Kwon GH … +2 more , Lee EH, Seo H
Comput Methods Programs Biomed · 2026 Jun · PMID 41895217 · Publisher ↗
BACKGROUND AND OBJECTIVE: Low-intensity transcranial focused ultrasound (LIFUS) has emerged as a promising approach for non-invasive brain stimulation and blood-brain barrier (BBB) opening in combination with microbubble...
BACKGROUND AND OBJECTIVE: Low-intensity transcranial focused ultrasound (LIFUS) has emerged as a promising approach for non-invasive brain stimulation and blood-brain barrier (BBB) opening in combination with microbubbles (MBs). Although ultrasound propagation in LIFUS and MB dynamics have been investigated in previous studies, they are often examined separately; thus, it remains challenging to establish safe and effective acoustic thresholds given the nonlinear nature of MB oscillations and distortions introduced by the skull. To address this gap, in this study, we developed a multi-scale computational model that integrates three-dimensional LIFUS simulations with MB dynamics modeled using the Marmottant equation. METHODS: We conducted LIFUS simulations using 3D human skull models derived from computed tomography and magnetic resonance imaging data of male and female subjects. Simulations were performed at two frequencies (250 and 500 kHz) with two transducer placements (back and upper side of the head). The resulting 3D acoustic pressure fields were subsequently coupled to the Marmottant equations, incorporating four clinically relevant microbubble sizes (0.82, 1.0, 1.75, and 2.15 μm), to quantify the spatial volumes of stable (V) and inertial cavitation (V). RESULTS: Both V and V increased with the acoustic amplitude and bubble radius but decreased with the frequency. Skull-induced distortions further enlarged the cavitation volumes relative to free water at the same peak pressure amplitude and temporally advanced the onset of cavitation by approximately 5 μs. When skull-induced attenuation was considered, acoustic pressures were reduced by 75%-85%, leading to smaller cavitation volumes at the same transducer output amplitude compared with free water. This attenuation effect also elevated the inertial cavitation threshold by approximately 3-5 fold. CONCLUSION: These findings highlight the critical role of skull anatomy in LIFUS and the effect of skull-induced distortion on MB cavitation. Using multi-scale modeling techniques, we obtained results similar to those of previous studies, thus providing a basis for safe and effective BBB opening and therapeutic applications.
A systematic review of generative artificial intelligence techniques for synthetic medical image datasets: Quality, models, public availability and applications.
Fu X, Li X, Delamare E … +3 more , Dunn AG, Bi L, Kim J
Comput Methods Programs Biomed · 2026 Jun · PMID 41895216 · Publisher ↗
BACKGROUND AND OBJECTIVE: Generative Artificial Intelligence (GAI) offers promising solutions to long-standing challenges in developing medical imaging methods and applications, including data scarcity, privacy concerns,...
BACKGROUND AND OBJECTIVE: Generative Artificial Intelligence (GAI) offers promising solutions to long-standing challenges in developing medical imaging methods and applications, including data scarcity, privacy concerns, and class imbalance. However, limited consolidation of publicly accessible synthetic datasets and trained GAI checkpoints restricts reproducibility and benchmarking. This systematic review aims to identify and evaluate such resources and assess their utility in clinical imaging applications. METHODS: We systematically searched PubMed, IEEE Xplore, and Scopus for studies published between January 2017 and June 2024. Eligible studies generated or used synthetic medical image datasets and publicly released either the dataset or the trained GAI model. Extracted data included imaging modality, dataset characteristics, model architecture, public availability, and evaluation strategy. RESULTS: Of 941 screened records, 35 studies met inclusion criteria, comprising 37 publicly available resources spanning radiology (59%), pathology (16%), ophthalmology (14%), and dermatology (11%). Generative models included generative adversarial networks (73%), diffusion models (21%), autoencoders (3%), and hybrid architectures (8%). As some studies employed multiple model types, these categories are not mutually exclusive. Fifteen (43%) studies provided trained model checkpoints, enabling the generation of task-specific synthetic data. Evaluation methods included quantitative metrics, clinical expert assessment, and downstream performance in classification, segmentation, or detection tasks. CONCLUSION: Although the reviewed resources support diverse downstream applications, publicly available synthetic datasets and trained models remain scarce. Evaluation strategies vary widely, and the absence of standardized benchmarks limits cross-study comparisons and reliability assessment. To support reproducibility and responsible use of GAI in medical imaging, future work should prioritize the public release of curated synthetic resources, clearer guidance on model selection, and standardized, multi-dimensional evaluation frameworks.
Identification of physiological parameters and missing inputs in glucose dynamics via PINN-constrained inference.
Zhang C, Ma Y
Comput Methods Programs Biomed · 2026 Jun · PMID 41886802 · Publisher ↗
BACKGROUND AND OBJECTIVE: Accurate glucose-insulin modeling under free-living conditions is challenged by incomplete or inaccurate input records, noisy continuous glucose monitoring data, and strong inter-individual phys...
BACKGROUND AND OBJECTIVE: Accurate glucose-insulin modeling under free-living conditions is challenged by incomplete or inaccurate input records, noisy continuous glucose monitoring data, and strong inter-individual physiological variability. These factors complicate reliable personalization and system identification. This study aims to develop a physiologically grounded identification framework capable of estimating subject-specific physiological parameters and unobserved exogenous inputs from partially observed data. METHODS: The proposed glucose latent input and parameter inversion (GLU-INVERT) framework extends the Bergman minimal model by incorporating additional physiological states and casting the identification task as a structured inverse problem with provable identifiability. A physics-informed learning mechanism embeds glucose and insulin dynamics as differentiable constraints, while an alternating optimization strategy estimates subject-specific physiological parameters and infers sparse latent meal correction signals. RESULTS: Parameter estimates obtained by the proposed framework remained within established physiological ranges and exhibited reduced inter-subject variability, indicating improved identifiability under incomplete input information. As a secondary validation, the identified model was evaluated in a rolling-horizon forecasting setting with fixed parameters. From the 30-minute to the 120-minute prediction horizon, the proposed GLU-INVERT framework achieved the lowest mean absolute relative difference (MARD), increasing moderately from 8.8% to 24.1%, whereas alternative approaches showed larger increases from 10.3% to 30.0%. Over the same horizons, GLU-INVERT also attained the lowest root mean squared error (RMSE), rising from 0.69 to 1.80,mmol/L, compared with increases from 0.78 to 2.29,mmol/L for the alternatives. Performance improvements over the least-squares baseline were statistically significant across all prediction horizons (p<0.05) and degraded more slowly with increasing horizon length, indicating enhanced stability under data-limited conditions. CONCLUSIONS: By addressing parameter uncertainty and missing input information, GLU-INVERT provides a robust and interpretable framework for physiological system identification under real-world data constraints. Forecasting performance is presented as a secondary validation of the identified model and highlights its potential utility for personalized glucose monitoring and decision support.
Prediction of femoral shaft fracture healing and clinical decision support through integration of mechano-biological modeling and machine learning.
Ni Z, Zhang J, Zhang P
Comput Methods Programs Biomed · 2026 Jun · PMID 41880959 · Publisher ↗
BACKGROUND AND OBJECTIVES: Current clinical practice in preoperative planning for femoral shaft fractures lacks tools capable of quantitatively predicting outcomes across different treatment options, which significantly...
BACKGROUND AND OBJECTIVES: Current clinical practice in preoperative planning for femoral shaft fractures lacks tools capable of quantitatively predicting outcomes across different treatment options, which significantly hinders the implementation of personalized precision treatment. This study aims to develop an integrated fracture healing prediction framework that combines mechano-biological modeling with machine learning. METHODS: First, a comprehensive mechano-biological model was constructed, incorporating four key modules: mechanical stimulus computation, angiogenesis prediction, cell migration and differentiation, and callus modulus updating, to dynamically simulate femoral shaft fracture healing under bone plate fixation. Subsequently, the model generated 729 datasets using bone plate modulus, fracture gap size, and loading conditions across four rehabilitation stages as input features, with cortical callus modulus at 16 weeks postoperation as the output target. Four machine learning algorithms-Support Vector Regression (SVR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN)-were systematically compared. RESULTS: The mechano-biological model demonstrated consistent trends with animal experimental data. XGBoost achieved optimal predictive performance (R² = 0.969, MSE = 0.045, RMSE = 0.211, MAE = 0.178, MAPE = 5.795%). Feature importance analysis revealed that bone plate modulus (28%) and fracture gap size (25%) were the most critical factors influencing healing quality, while Stage 3 loading (weeks 9-12 postoperation, 18%) represented a critical window for mechanical intervention. CONCLUSIONS: Optimizing implant stiffness and mechanical stimulation during critical phases can effectively improve healing outcomes. The integrated framework provides a reliable theoretical tool to support personalized clinical decision-making.
dMMR prediction from colorectal cancer histopathology: Leveraging non-tumor and low-magnification regions.
Petäinen L, Väyrynen JP, Böhm J … +13 more , Ruusuvuori P, Ahtiainen M, Elomaa H, Karjalainen H, Kastinen M, Tapiainen VV, Äijälä VK, Sirniö P, Tuomisto A, Mäkinen MJ, Mecklin JP, Pölönen I, Äyrämö S
Comput Methods Programs Biomed · 2026 Jun · PMID 41875848 · Publisher ↗
BACKGROUND AND OBJECTIVE: Colorectal cancer is the second leading cause of cancer-related mortality worldwide, posing a substantial burden on healthcare systems. Identifying DNA mismatch repair deficiency (dMMR) is criti...
BACKGROUND AND OBJECTIVE: Colorectal cancer is the second leading cause of cancer-related mortality worldwide, posing a substantial burden on healthcare systems. Identifying DNA mismatch repair deficiency (dMMR) is critical for guiding treatment, yet conventional methods rely on labor-intensive DNA analysis. While deep-learning approaches have shown promise for predicting dMMR from histopathological images, most studies focus exclusively on tumor regions and single-scale representations. This study systematically evaluates the predictive value of tumor and non-tumor regions across multiple magnifications for dMMR prediction from whole-slide images (WSIs). METHODS: A total of 24 different modeling approaches were evaluated, varying by tissue origin (tumor vs. non-tumor), magnification level (5x and 20x), and tile embedding strategy, including digital pathology foundation models. Tile embeddings were further trained with 1228 WSIs using multiple-instance learning (MIL) based approach. The best-performing configurations were selected for external evaluation. External testing was carried out on two independent cohorts consisting of 1010 and 457 WSIs, respectively. RESULTS: Non-tumorous regions demonstrated measurable predictive value, although performance remained lower than that obtained from tumor regions (F1 = 0.896, precision = 0.888, sensitivity = 0.594, specificity = 0.982). Among the nine models selected during internal validation, the top three models-one multi-scale approach and two models trained on 20x tumor regions-achieved F1 scores of 0.870-0.889 with precision of 0.885-0.920, sensitivity of 0.852, and specificity of 0.889-0.926. On external validation, the top three models, all based on foundation-model tile embeddings, achieved F1 scores of 0.916-0.919 on the first cohort and 0.928-0.934 on the second cohort. Across cohorts, specificity remained consistently high (0.964-0.992), while sensitivity ranged from 0.500 to 0.682. CONCLUSION: This study demonstrates that dMMR status in colorectal cancer can be effectively predicted from histopathological WSIs using MIL-based models, with moderate generalizability across independent cohorts. In addition to confirming the predictive value of tumor regions, the results reveal that non-tumorous tissue also contains detectable predictive signals, suggesting that microenvironmental features may contribute to dMMR-associated histological patterns. Furthermore, the use of foundation model-derived embeddings improved generalizability across datasets. Future work should explore integrating non-tumor tissue features and clinical data to further improve predictive performance.
A hybrid implicit-explicit finite element framework for real-time bioheat transfer simulation in deformable biological tissue.
Wang F, Liu PX
Comput Methods Programs Biomed · 2026 Jun · PMID 41875847 · Publisher ↗
BACKGROUND AND OBJECTIVE: Real-time simulation of bioheat transfer in deformable tissues is essential for realistic surgical training, yet it remains challenging due to stringent requirements for numerical stability and...
BACKGROUND AND OBJECTIVE: Real-time simulation of bioheat transfer in deformable tissues is essential for realistic surgical training, yet it remains challenging due to stringent requirements for numerical stability and computational efficiency. To overcome these limitations, we propose a unified finite element framework that seamlessly integrates implicit and explicit schemes, enabling real-time updates of tissue deformation while maintaining computationally efficient thermal simulations. METHODS: This paper proposes a novel hybrid finite element framework that employs an optimization-based implicit time integration scheme for tissue mechanics, ensuring numerical stability even under large deformations, while utilizing an explicit time-integration scheme for the Pennes bioheat transfer model to achieve computationally efficient thermal simulations. Additionally, the framework integrates a physiological motion model to reproduce realistic tissue dynamics, enhancing the fidelity of surgical simulation. RESULTS: Validation against commercial software Abaqus and COMSOL under pure conduction, blood perfusion, and motion scenarios demonstrates excellent accuracy, with maximum normalized relative error below 0.4%, RMSE below 0.009 °C, and RLNE below 0.0015 across all scenarios. GPU-accelerated thermal computation achieved single-step execution times below 50μs for meshes up to 50,000 elements. Real-time performance was confirmed on consumer-grade hardware in liver ablation simulations, highlighting the framework's suitability for interactive surgical training applications. CONCLUSION: The hybrid implicit-explicit strategy effectively balances numerical stability with computational efficiency in coupled thermo-mechanical simulations. The demonstrated accuracy and real-time performance highlight the framework's potential for interactive surgical training applications, particularly in thermal ablation therapy.
MOSAIC: Multi-scale orientation-aware segmentation and instance classification network for histopathological image analysis.
Wadood AS, Fauzi MFA, Wong LK … +3 more , Lee JTH, Khor SY, Looi LM
Comput Methods Programs Biomed · 2026 Jun · PMID 41871485 · Publisher ↗
BACKGROUND AND OBJECTIVE: Accurate nuclei segmentation and instance classification are fundamental tasks in biomedical image analysis; however, many existing computational models exhibit limited robustness when confronte...
BACKGROUND AND OBJECTIVE: Accurate nuclei segmentation and instance classification are fundamental tasks in biomedical image analysis; however, many existing computational models exhibit limited robustness when confronted with scale variability, morphological heterogeneity, and arbitrary rotational orientations commonly observed in histopathological images. The objective of this work is to develop a unified computational framework that is robust to effective magnification variability, arbitrary orientations, and long-range contextual dependencies, without relying on multi-magnification supervision or magnification-specific retraining. METHODS: We propose a multi-scale orientation-aware segmentation and instance classification (MOSAIC) framework, which integrates hierarchical context extraction, rotation-aware feature fusion, and transformer-based long-range contextual modeling within a single encoder-decoder architecture. The proposed model combines large-, medium-, and small-scale contextual cues derived from a single native training magnification to enable robust learning across effective magnifications. The proposed method is evaluated on an institutional estrogen receptor immunohistochemistry cohort, the multi-organ nuclei segmentation and classification dataset, and the colorectal nuclei segmentation and phenotypes dataset. RESULTS: The proposed model outperforms baseline methods, achieving a mean Dice coefficient of 0.862, an Aggregated Jaccard Index of 0.721, and a Panoptic Quality score of 0.647, with consistent improvements of 3%-7% across datasets. The model also demonstrates favorable computational cost relative to representative baselines, with an inference time of 0.175 s per 512 × 512 image patch and a peak memory footprint of 3.7 GB. CONCLUSIONS: The results demonstrate that orientation-aware multi-scale fusion and long-range contextual modeling improve boundary precision, instance separation, and classification consistency across heterogeneous nuclear morphologies. These improvements indicate that the proposed design generalizes reliably across challenging tissue appearances.
PIBAdb: a public cohort of multimodal colonoscopy videos and images including polyps with histological information.
Nogueira-Rodríguez A, Domínguez-Carbajales R, Campos-Tato F … +13 more , Herrero J, Puga M, Rivas L, Sánchez E, Iglesias Á, Codesido L, Díez-Martín ÁI, Davila-Piñón P, Cubiella J, Fdez-Riverola F, López-Fernández H, Glez-Peña D, Reboiro-Jato M
Comput Methods Programs Biomed · 2026 Jun · PMID 41863886 · Publisher ↗
BACKGROUND AND OBJECTIVE: Colorectal cancer is the third most common cancer worldwide and presents a high mortality rate. Colonoscopy is the gold standard for screening, as it can reduce its incidence and mortality. Deep...
BACKGROUND AND OBJECTIVE: Colorectal cancer is the third most common cancer worldwide and presents a high mortality rate. Colonoscopy is the gold standard for screening, as it can reduce its incidence and mortality. Deep Learning techniques have become state-of-the-art in lesion detection and classification, and several Deep-Learning-based Computer-Aided Diagnosis systems are already undergoing clinical evaluation or commercialization. However, the development of reliable models requires large, high-quality datasets, which are costly and time-consuming to create. Thus, the availability of public datasets is critical for the scientific community to develop artificial intelligent models. This work aims to contribute to the available resources by presenting PIBAdb, a new multimodal public cohort of colorectal videos and images. METHODS: The PIBAdb cohort contains polyp data derived from routine colonoscopies conducted between January 2018 and May 2021 at Hospital Universitario de Ourense, under the PolyDeep project. Each polyp was resected, histologically analysed, morphologically classified, and annotated by expert clinicians with bounding boxes in images and temporal segments in videos. The main characteristics of PIBAdb were compared with another 25 public datasets. The utility of PIBAdb was evaluated in polyp detection and classification scenarios using Deep Learning models. RESULTS: PIBAdb includes detailed clinical and histological metadata from 1176 polyps, 31,946 manually annotated polyp images, 14,124 non-polyp images, nearly 7 h of annotated video segments showing polyps, and over 4 h of annotated video segments without polyps. It comprises both the raw database and several curated image datasets, each accompanied by metadata and documentation. PIBAdb is publicly available upon request for non-profit purposes. CONCLUSIONS: PIBAdb is one of the largest and most complete multimodal public datasets for colorectal polyp research. It is characterized by its rich per-polyp metadata (histology and PARIS/NICE classifications), inclusion of NBI and WL images, and non-polyp images at multiple levels of cleanness. While the image datasets included are practical for developing classification or detection models, the full database enables more complex video-based research and custom dataset creation using the PIBA management tool, supported by a queryable relational database. Its availability is expected to support the development of Deep Learning models and to foster future contributions from the research community.
Non-invasive characterization of tumors using Bayesian inference and Virtual Element Method.
Thananjayan P, Arrutselvi M, Natarajan S
Comput Methods Programs Biomed · 2026 Jun · PMID 41861422 · Publisher ↗
BACKGROUND: Accurate and non-invasive detection of tumor cells remains a major challenge in biomedical engineering and clinical diagnostics. Traditional imaging methods often face limitations in resolution, accessibility...
BACKGROUND: Accurate and non-invasive detection of tumor cells remains a major challenge in biomedical engineering and clinical diagnostics. Traditional imaging methods often face limitations in resolution, accessibility, or invasiveness. We propose a computational framework combining Bayesian inference with the Virtual Element Method (VEM) to address the inverse problem of tumor characterization using surface temperature measurements. METHODS: The forward thermal response of biological tissues was modeled using Pennes' bioheat equation, with skin surface temperature distributions as measurable data. Three test scenarios were designed: (1) detecting and quantifying a single, small, elliptical tumor using the Metropolis-Hastings (M-H) algorithm, (2) identification of a cluster of non-elliptical-shaped fragments using M-H algorithm and (3) simultaneous estimation of the number, locations, and sizes of multiple tumors using the Reversible Jump Markov Chain Monte Carlo (RJMCMC) algorithm; and assessing the robustness of both inference strategies under varying levels of simulated measurement noise. RESULTS: In scenarios 1 and 2, the M-H algorithm successfully detected and quantified the single tumor and cluster of tumor cells, demonstrating reliability for localized anomalies. In scenario 3, the RJMCMC algorithm accurately estimated multiple tumor parameters simultaneously, demonstrating the framework's capability to address complex multi-tumor scenarios. Both inference approaches exhibited strong robustness across varying noise levels, ensuring reliable tumor detection and characterization under modeling and measurement noise. CONCLUSION: The integration of Bayesian inference with the VEM provides a flexible and powerful computational framework for non-invasive tumor detection and characterization. This approach shows strong potential for enhancing thermal-based tumor detection by offering improved reliability and adaptability for clinical diagnostics. Moreover, unlike traditional optimization-based inverse methods, which provide only point estimates, the proposed Bayesian framework yields credible intervals for all inferred parameters, enabling uncertainty quantification, particularly valuable for clinical interpretation.
Med-ViX-Ray: Enhancing explainable chest X-ray analysis with clinical knowledge graphs.
Cieri M, Palomba F
Comput Methods Programs Biomed · 2026 Jun · PMID 41850037 · Publisher ↗
BACKGROUND AND OBJECTIVE: Deep learning has achieved remarkable success in chest x-ray interpretation, yet most models remain black boxes, producing accurate predictions without exposing the clinical reasoning behind the...
BACKGROUND AND OBJECTIVE: Deep learning has achieved remarkable success in chest x-ray interpretation, yet most models remain black boxes, producing accurate predictions without exposing the clinical reasoning behind them. This opacity limits trust and adoption in real-world practice. We introduce Med-ViX-Ray, a knowledge-guided and interpretable framework that integrates symbolic clinical reasoning into a vision Transformer backbone. METHODS: The model leverages a structured graph of radiological signs and conditions, aligning image attention maps with domain knowledge through a probabilistic soft-matching module and a nudging mechanism that refines classifier outputs. This dual integration allows predictions to be explained in terms of clinically meaningful signs and corresponding image regions, offering transparency beyond post-hoc heatmaps. We evaluated Med-ViX-Ray on MIMIC-CXR for training and internal validation, and tested its generalization on VinDR-CXR and RSNA Pneumonia benchmarks. RESULTS: The proposed method improves recall and F1-score compared to a strong SwinV2 baseline (Respectively, F1-micro: 0.561 - 0.456; Precision: 0.462 - 0-529; Recall: 0.715 - 0.466; ROC: 0.788 - 0.744), while maintaining competitive overall performance. Qualitative analyses confirm that the model highlights clinically relevant regions and sign-activations aligned with radiological practice. CONCLUSION: These results suggest that knowledge-guided attention and sign-based explanations can enhance interpretability and recall in chest X-ray classification models. Future work will extend the framework toward report generation and prospective clinical evaluation.
A robust topology optimization based biomechanical computational framework for patient-specific trabecular bone microstructure reconstruction.
Li Z, Huang X, Ding Z … +4 more , Featherston C, Evans SL, Zioupos P, Wu Z
Comput Methods Programs Biomed · 2026 Jun · PMID 41850036 · Publisher ↗
BACKGROUND AND OBJECTIVE: Accurate reconstruction of trabecular bone microstructure is essential for understanding bone health and mechanical competence. Low-resolution computed tomography images, however, lack the detai...
BACKGROUND AND OBJECTIVE: Accurate reconstruction of trabecular bone microstructure is essential for understanding bone health and mechanical competence. Low-resolution computed tomography images, however, lack the detailed information that is needed to depict fine trabecular architecture. This study aims to develop a computational framework that reconstructs subject-specific trabecular microstructure with improved accuracy and stability by incorporating mechanical and biological variability inherent in bone adaptation. METHODS: A robust topology optimization framework was developed to predict trabecular morphology from low-resolution images. The method incorporates uncertainty in loading and biological response during bone remodeling. To reduce sensitivity to variations in boundary forces, a superposition strategy was used to estimate local mechanical stimuli within each volume of interest. The predicted microstructure was compared against high-resolution images of rabbit bone for validation, and subsequently applied to human lower-limb bone images. Quantitative assessments included geometric similarity and evaluation of mechanical anisotropy. RESULTS: The reconstructed trabecular regions showed close agreement with high-resolution microstructural images in the animal validation study, capturing fine branching and connectivity patterns. In human bone, the predicted morphology was consistent with expected statistical distributions of trabecular thickness, spacing, and orientation. The framework demonstrated high computational precision and stability, producing anisotropic mechanical properties aligned with physiological loading patterns. CONCLUSIONS: This computational approach enables patient-specific reconstruction of trabecular microstructure from low-resolution imaging with improved robustness and reduced computational cost. The framework shows potential for supporting clinical assessment and for advancing multi-scale investigations of bone mechanics.
CRMIPred: Identifying the spatial interactions among cis-regulatory modules via considering their cross-attended epigenetic profiles.
Yu YH, Dai WC, Jiang ZH … +1 more , Yang TH
Comput Methods Programs Biomed · 2026 Jun · PMID 41833256 · Publisher ↗
BACKGROUND AND OBJECTIVE: Modular DNA elements known as cis-regulatory modules (CRMs) play central roles in transcriptional regulation in metazoan species. Beyond their individual functions, CRMs can physically interact...
BACKGROUND AND OBJECTIVE: Modular DNA elements known as cis-regulatory modules (CRMs) play central roles in transcriptional regulation in metazoan species. Beyond their individual functions, CRMs can physically interact with one another to cooperatively regulate target gene expression, forming an additional layer of transcription regulation. METHODS: Experimental identification of such interactions typically relies on chromosome conformation capture technologies coupled with sequencing (HiC), which require high sequencing depths to achieve sufficient resolution for CRM-level analysis, resulting in substantial cost. Computational approaches, therefore, provide an economic strategy for pre-screening potential CRM interactions. Nonetheless, existing tools often lack sufficient resolution and are restricted to limited CRM types. Some tools even suffer from data contamination caused by improper data partitioning. Here, we presented CRMIPred (CRM Interaction Predictor), a deep learning framework for CRM interaction identification built upon a chromosome-based, data-snooping-free partitioning scheme. CRMIPred models epigenetic crosstalk between CRMs using a cross-attention architecture to capture biologically meaningful interactions between multi-track epigenetic profiles. RESULTS: On a strictly held-out test set, CRMIPred achieved an auROC of 87.7% and an auPRC of 89.3% in recognizing interacting CRM pairs, outperforming all currently available tools and baseline methods by over 10.3% and 7.2% in auROC and auPRC, respectively. Moreover, the model demonstrated robustness to input design choices, and further analyses confirmed that its performance gains stem from the biologically grounded cross-attention mechanism. CONCLUSIONS: Beyond its use as a pre-screening tool, CRMIPred also provides a computational framework for investigating the mechanistic relationship between epigenetic codes and chromatin interactions, offering insight into how epigenetic crosstalk mediates CRM-CRM communication. CRMIPred is available at https://github.com/cobisLab/CRMIPred.
Inhalation exposure and particle deposition across 16 nonhuman primate airway models using computational fluid-particle dynamics for allometric extrapolation.
Suda Y, Kuga K, Khoa ND … +2 more , Nishimura T, Ito K
Comput Methods Programs Biomed · 2026 Jun · PMID 41830842 · Publisher ↗
BACKGROUND AND OBJECTIVE: Nonhuman primates are indispensable surrogate models for humans due to their evolutionary proximity; however, ethical concerns and practical constraints are increasingly limiting their use. Comp...
BACKGROUND AND OBJECTIVE: Nonhuman primates are indispensable surrogate models for humans due to their evolutionary proximity; however, ethical concerns and practical constraints are increasingly limiting their use. Computational fluid-particle dynamics (CFPD) offers a promising alternative for inhalation exposure studies; however, previous studies typically used small sample sizes with limited model validation. Therefore, this study aims to elucidate inhalation exposure in monkeys through a large cohort analysis and to identify relationships between humans and monkeys by allometric comparison. METHODS: In this study, we reconstruct anatomically accurate upper airway models from computed tomography (CT) data of 16 macaques (Macaca fuscata and Macaca mulatta). Steady-state airflow and particle transport simulations are performed using high-resolution meshes with verified grid independence. Particle aspiration and deposition efficiencies are evaluated across 11 microparticles using the Lagrangian particle tracking. RESULTS: Airway geometry was found to strongly influence the airflow resistance, velocity distribution, and wall shear stress. These factors resulted in substantial interindividual variability in aspiration and particle deposition patterns. Comparative analysis with pediatric human models revealed morphological correspondences. The mean nasal cross-sectional area correlated more strongly with age than the minimal nasal cross-sectional area or region-specific areas. Normalizing deposition efficiency by a modified Stokes number substantially reduced intersubject variability and yielded a high-fidelity logistic fit. CONCLUSIONS: These findings emphasize the importance of incorporating both age- and morphology-dependent factors into interspecies extrapolations. Therefore, this study provides a basis for more reliable assessments of inhalation exposure and enhances the validity of extrapolating data from monkeys to humans.
Robust prediction of parameterized cardiovascular hemodynamics using deep operator networks with time normalization.
Hong J, Lee B, Persad AR … +5 more , Song H, Min C, Jeong JH, Doostan A, Park YH
Comput Methods Programs Biomed · 2026 Jun · PMID 41830841 · Publisher ↗
BACKGROUND AND OBJECTIVES: Computational modeling of cardiovascular hemodynamics is essential for understanding disease mechanisms. While one-dimensional (1D) numerical solvers are widely used, their cumulative computati...
BACKGROUND AND OBJECTIVES: Computational modeling of cardiovascular hemodynamics is essential for understanding disease mechanisms. While one-dimensional (1D) numerical solvers are widely used, their cumulative computational cost becomes prohibitive in many-query scenarios - such as global sensitivity analysis and Bayesian parameter estimation - that require thousands of iterative evaluations. This study presents a parameterized Deep Operator Network (DeepONet) framework integrated with time normalization to enable rapid and robust surrogate modeling. METHODS: The proposed framework employs a two-step time normalization strategy comprising cycle and period normalization. A parameterized trunk network explicitly encodes heart rate, arterial stiffness, and relative arterial length to compensate for temporal scaling and facilitate explicit parameterization. The method was evaluated on simulated 1D cardiovascular cases using a testing protocol that included out-of-distribution scenarios with parameters extending significantly beyond the training range. RESULTS: The model achieved high accuracy within the training distribution, with a root mean squared error (RMSE) below 0.5 mmHg. In out-of-distribution regimes where standard models showed performance degradation (RMSE >20 mmHg), the proposed framework maintained relatively robust performance (mean RMSE <7 mmHg). Inference time was approximately 0.00038 s per waveform, representing a computational acceleration of around 150,000 times compared to numerical solvers. The framework's generalizability was further corroborated on a periodic structural dynamics problem predicting cantilever beam displacements. CONCLUSIONS: These results demonstrate that integrating time normalization with parameterized operator learning enables robust modeling of periodic physiological signals. This approach provides a computational foundation for solving computationally intensive inverse problems and performing large-scale sensitivity analyses, where standard numerical simulations are practically infeasible.
Dual modal data-driven virtual reality-based mild cognitive Impairment assessment using MCIformer.
Zhang Y, Pan Y, Feng S … +3 more , Qiao H, Bu L, Li F
Comput Methods Programs Biomed · 2026 May · PMID 41825293 · Publisher ↗
BACKGROUND: Mild Cognitive Impairment (MCI) assessment is critical for identifying cognitive decline and enabling early intervention to reduce the risk of Alzheimer's disease (AD). Virtual Reality (VR)-based cognitive as...
BACKGROUND: Mild Cognitive Impairment (MCI) assessment is critical for identifying cognitive decline and enabling early intervention to reduce the risk of Alzheimer's disease (AD). Virtual Reality (VR)-based cognitive assessments offer enhanced engagement, ecological validity, and user-friendliness. However, most existing VR-based systems rely primarily on task performance metrics, overlooking subtle changes in motor behaviour and underlying neural activity patterns that are indicative of early cognitive decline. METHODS: We developed a dual-modal, data-driven VR-based MCI assessment method integrating kinematic and functional near-infrared spectroscopy (fNIRS) data. A VR-based experiment involving healthy and MCI participants was conducted to collect synchronised kinematic and neural data. Kinematic features-smoothness, coordination, and stability-were extracted from segmented movement trajectories and structured as time series to capture subtle motor deficits. Simultaneously, multichannel fNIRS data were represented as functional brain networks to assess interregional connectivity changes associated with cognitive impairment. We proposed MCIformer, a dual-modal fusion model that applies a Transformer to kinematic sequences to capture dynamic motor patterns and a Graph Transformer to fNIRS networks to detect connectivity alterations. RESULTS: Experimental results show that the proposed system achieved 90% accuracy, significantly outperforming models using only kinematic data (80%) or only fNIRS data (85%). The findings demonstrate that integrating temporal motor patterns and spatial brain connectivity patterns enhances classification performance by capturing complementary brain-behaviour information. CONCLUSIONS: The proposed VR-based dual-modal MCI assessment approach demonstrates strong potential for scalable, accurate early diagnosis in community settings, supporting the development of more comprehensive brain-behaviour monitoring systems for cognitive health.
A deep clustering Gaussian process algorithm for motor progression prediction of Parkinson's disease.
Pan C, Tian Y, Zhou T … +6 more , Chi S, Wu C, Ma L, Ren K, Feng T, Li J
Comput Methods Programs Biomed · 2026 May · PMID 41825292 · Publisher ↗
BACKGROUND AND OBJECTIVE: Parkinson's disease (PD) is a chronic progressive neurodegenerative disorder characterized by significant spatial and temporal heterogeneity in symptom presentation and progression, which poses...
BACKGROUND AND OBJECTIVE: Parkinson's disease (PD) is a chronic progressive neurodegenerative disorder characterized by significant spatial and temporal heterogeneity in symptom presentation and progression, which poses a major challenge for accurate motor progression prediction. Developing a highly individualized model for predicting PD motor progression that can reflect the heterogeneity may lead to better management. METHODS: To address this issue, we propose a novel Deep Clustering Gaussian process (DCGP) algorithm to predict PD motor progression using readily available clinical statistics and scales. The algorithm consists of three modules: a pretraining module to obtain valuable latent representations, a clustering module to capture heterogeneity among different progression patterns, and an adaptive module to fine-tune models for specific patient clusters. RESULTS: Experiments conducted on 342 patients from the Parkinson's Progression Markers Initiative (PPMI) and 336 patients from the National Institute of Neurological Disorders and Stroke Parkinson's Disease Biomarkers Program (PDBP) demonstrated that our proposed DCGP significantly outperforms existing state-of-the-art predictive models (PPMI: RMSE = 4.607 ± 0.218, MAE = 3.504 ± 0.138, R² = 0.890 ± 0.012; PDBP: RMSE = 7.486 ± 0.208, MAE = 5.477 ± 0.157, R² = 0.652 ± 0.020). The ablation experiments show that our proposed algorithm improves predictive performance by capturing heterogeneity among different progression patterns. Based on the proposed DCGP, the magnitude of this disease progression heterogeneity was quantified as the difference between average levels and the variation over time and results reveal that the PDBP cohort exhibits greater heterogeneity in average disease levels, whereas the PPMI cohort shows greater heterogeneity in progression rate, trend, and smoothness. CONCLUSIONS: This study proposes a novel DCGP algorithm to predict PD motor progression, which can enhance predictive performance by quantitatively capturing heterogeneity, thereby aiding doctors in making accurate predictions and providing tailored management plans for PD patients.
Geometry-aware PointNet for rapid prediction of cerebral aneurysm hemodynamics.
Sheng Y, Liao C, Li W … +4 more , Yang E, Zhu Y, Sun H, Leo HL
Comput Methods Programs Biomed · 2026 May · PMID 41825291 · Publisher ↗
BACKGROUND AND OBJECTIVE: Cerebral aneurysms affect 2-5% of the global population and pose a significant health risk upon rupture. While computational fluid dynamics provides detailed hemodynamic information for risk ass...
BACKGROUND AND OBJECTIVE: Cerebral aneurysms affect 2-5% of the global population and pose a significant health risk upon rupture. While computational fluid dynamics provides detailed hemodynamic information for risk assessment, its high computational demands limit routine clinical use. This study aims to develop a deep learning model to rapidly predict three-dimensional velocity fields and wall shear stress at peak systole using aneurysm geometry. METHODS: We synthesized a dataset of 984 idealized middle cerebral artery bifurcation aneurysms. For each case, computational fluid dynamics simulations were conducted with pulsatile boundary conditions to generate ground-truth data, and peak-systolic snapshots were extracted. We developed a single-input point-cloud network augmented with a distance-to-wall feature to predict both velocity fields and wall shear stress. Model performance was evaluated using mean absolute error (MAE), normalized MAE (NMAE), and relative L2 error (rL2). RESULTS: On the test set, the model achieved velocity-field accuracy of NMAE 4.05% and rL2 19.2% over the full geometry domain, and WSS accuracy of NMAE 2.59% and rL2 23.9%. The mean inference time was approximately 1.6 seconds for velocity and 0.3 seconds for wall shear stress per case. Out-of-distribution evaluation on non-idealized geometries showed substantial zero-shot degradation (NMAE 19.1%, rL2 62.4%), while leave-one-out fine-tuning improved performance (NMAE 10.8%, rL2 37.0%). CONCLUSIONS: The proposed geometry-aware point-cloud surrogate provides fast peak-systolic hemodynamic prediction on idealized aneurysm geometries. However, out-of-distribution evaluation indicates that broader patient-specific training, physiological boundary conditions, and reliability assessment are required before routine clinical application.
Graph empirical mode decomposition and multiscale feature extraction for EEG-based classification of Alzheimer's disease and frontotemporal dementia.
Acharya M, Deo RC, Barua PD … +2 more , Devi A, Tao X
Comput Methods Programs Biomed · 2026 May · PMID 41806548 · Publisher ↗
BACKGROUND AND OBJECTIVE: Early and correct classification of neurodegenerative diseases like Alzheimer's Disease (AD) and Frontotemporal Dementia (FTD) is one of the most important challenges in clinical neurology. In t...
BACKGROUND AND OBJECTIVE: Early and correct classification of neurodegenerative diseases like Alzheimer's Disease (AD) and Frontotemporal Dementia (FTD) is one of the most important challenges in clinical neurology. In this paper, we present a novel electroencephalogram (EEG)-based approach that integrates a rich set of multiresolution features to improve the performance of automatic classification. METHOD: Our approach fuses the Graph Fourier Transform (GFT), Graph Wavelet Transform (GWT), Discrete Wavelet Transform (DWT), and a newly developed Graph Empirical Mode Decomposition (GEMD) technique to primarily boost the performance of the proposed model. This also retained the complementary spatial, spectral, and temporal information carried by the EEG signals, which are significant for the differentiation of AD, FTD, and HC subjects. The EEG recordings were segmented into fixed lengths with non-overlapping windows of four durations: 1000, 5000, 10,000, and 20,000 samples. Energy and entropy features were obtained for each segment, both individually within domains and combined into a single 388-dimensional feature vector. The features were then normalized and fed into various machine learning (ML) models, including support vector machines (SVMs), k-nearest neighbors (kNNs), decision trees (DTs), random forests (RFs), and an ensemble learning model with the AdaBoost capability. RESULTS: The proposed model was tested using accuracy, precision, recall, specificity, and F1-scores, with results showing that the ensemble model was better than the other benchmark models in every classification task. That is, in this binary classification problem, an accuracy of 98.84% for AD vs. HC, 98.67% for AD vs. FTD, and 98.94% for FTD vs. HC was obtained. CONCLUSION: In the multiclass task (AD, FTD, HC), the method reached 96.68% accuracy, demonstrating the efficacy of the proposed method for the identification of Alzheimer's disease and frontotemporal dementia. Compared to previous research using the same dataset, our approach has demonstrated improved performance, validating the effectiveness of graph-based multiresolution feature fusion for dementia classification using EEG signals.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe