Annals of Biomedical Engineering最新文献

Purpose: Constipation is a prevalent disorder of the digestive system, and abdominal massage is commonly employed to alleviate its symptoms due to its non-invasive nature and minimal side effects. This study aims to develop an accurate correlation model that links massage force (MF), massage depth (MD), and intra-abdominal wall deformation (IAWD) based on patient-specific data, thereby providing quantitative standards and personalized treatment options for the use of clinical massage devices.

Methods: The study utilized finite element method (FEM) and machine learning (ML) techniques to construct a digital twin model of the abdomen. Initially, a high-fidelity skin-muscle two-layer abdominal finite element model was developed using computed tomography (CT) images and mechanical experimental data. Subsequently, four individual learning models were trained, and stacked models were constructed by combining the advantages of these individual models through random forest and K-nearest neighbor algorithms to predict MF and IAWD.

Results: The prediction accuracy rates for MF in the ascending colon, transverse colon, and descending colon regions were 84.80%, 84.00%, and 85.60%, respectively. The prediction accuracy rates for IAWD were 96.00%, 87.00%, and 95.20%, respectively.

Conclusion: Compared to traditional finite element simulations, this abdominal digital twin model is capable of delivering predictions within 3 seconds, offering valuable insights for the digital transformation of healthcare.

Purpose: Stroke is a leading cause of death worldwide, with atrial fibrillation (AF) contributing to up to a third of ischemic strokes. Current clinical risk scores, like the CHA $_2^{}$ DS $_2^{}$ -VASc score, have limited accuracy and do not account for the underlying mechanisms of thrombus formation and stroke. This study investigates the potential of supplementing the CHA $_2^{}$ DS $_2^{}$ -VASc score with information about the form and function of the left atrium (LA) using computational fluid dynamics (CFD)-based flow parameters and morphometrics.

Methods: We analyzed 40 patient-specific LA models reconstructed from computed tomography, incorporating a physiologically realistic atrial wall motion derived from an AF motion model. High-fidelity CFD simulations were performed under pathological conditions to compute hemodynamic and morphometric parameters, which were then statistically correlated with thrombus presence or stroke history.

Results: Univariate logistic analysis showed that LA volume, left atrial appendage (LAA) orifice size, and LAA depth correlated strongly with thrombus presence or stroke history (Pseudo $R^2$ = 0.57, 0.55, 0.47, respectively). Additionally, increased LA blood residence time (BRT) showed a strong correlation with thrombus or stroke (Pseudo $R^2$ = 0.61). When combined with CHA $_2^{}$ DS $_2^{}$ -VASc  multivariate logistic regression produced strong Pseudo $R^2$ scores of 0.75, 0.75, and 0.76 for LAA orifice size, LAA depth, and LA BRT, respectively.

Conclusion: These results demonstrate that CFD-based and morphometric parameters can complement the CHA $_2^{}$ DS $_2^{}$ -VASc score, potentially improving the assessment of thrombus formation and stroke risk in AF patients, which could support more personalized treatment options.

Purpose: Clinically significant bleeding is a common complication of extracorporeal membrane oxygenation (ECMO), with bleeding observed in upwards of 40% of patients. Bleeding is strongly associated with systemic anticoagulation, which is typically necessary to reduce the incidence of thrombosis in the extracorporeal circuit. The InFlo MOBYBOX integrates a multimodal strategy for coagulation mitigation, combining optimized circuit components (pump, oxygenator) design with a biopassive hydrophilic surface coating to address both flow-related and surface-mediated contributors to thrombosis and hemolysis. This study investigated whether such a design could reduce thrombotic burden under conditions typically associated with high coagulation risk, low blood flow rates, thereby potentially reducing reliance on systemic anticoagulation.

Methods: We conducted a seven-day ovine study (n = 4) with InFlo MOBYBOX ECMO systems coated tip-to-tip with a biopassive hydrophilic coating under continuous low-flow conditions (1.8-2.2 L/min) without systemic anticoagulants. ECMO oxygen delivery and carbon dioxide removal were measured across the duration of the run, and clotting, hematology, and organ damage were observed at the end of therapy.

Results: The InFlo MOBYBOX exhibited 2.3 ± 3.1% clot coverage localized to the inlet helical pathway, with no device degradation, systemic thrombosis, or end-organ damage. Hematologic indices remained stable, plasma-free hemoglobin confirmed the absence of hemolysis, and oxygen transfer performance was preserved throughout support.

Conclusion: These preliminary findings suggest that the InFlo MOBYBOX may enable anticoagulant-free low-flow support with minimal thrombosis, hemolysis, and no organ injury, demonstrating the potential value of a multimodal approach using optimized fluidic pathways and a proprietary biopassive hydrophilic surface coating to advance safer extracorporeal therapies.

Purpose: Reinforcement learning with reversal is a paradigm frequently employed in neuropsychology to assess the adaptive capacity and flexibility in individuals in non-stationary environments. Each single behavior, however, depends on the superimposition of multiple factors, which are still not completely understood.

Methods: We optimize a neurocomputational model of the Basal Ganglia to simulate the behavior of 18 patients with Parkinson's disease (both ON and OFF medication) and 14 control subjects during a two-choice probabilistic reversal learning task (with 80-20% reward probabilities). The individual behavior (in terms of the cumulative number of correct responses during a 40-trials direct phase and a 40-trial reversal phase) is reproduced by fitting a few model parameters for each individual, representing the tonic dopamine level, Hebbian learning, and the exploratory attitude (noise level).

Results: Results show that very different responses can be explained quite well, ascribing them to the varying combination of the aforementioned individual factors. The tonic dopamine level is significantly different for patients in ON and OFF medication, while the other parameters were not statistically different. A regression analysis reveals that the value of the Hebbian learning rate is correlated with the subject's sensitivity to punishments. In contrast, the noise standard deviation is correlated with exploration, i.e., the tendency to modify the choice even after a reward.

Conclusion: The results provide a mechanistic explanation of the various factors that affect adaptation and flexibility in reinforcement learning, representing a first step toward characterizing and understanding the diverse behaviors on an individual basis.

Purpose: Characterization of anisotropic nonlinear material properties along with fiber directions for soft tissues based on known deformations (strains) and external loads is a significant clinical goal.

Methods: A gradient-based inverse solver is developed to retrieve anisotropic nonlinear tissue properties. By directly minimizing nodal force residuals using known deformed and reference configurations, the method avoids repeated forward simulations. To represent complex, non-trivial spatial distributions, the material properties are parameterized using a multilayer perceptron (MLP) that maps spatial location to material behavior, in contrast to methods that use such networks to approximate the displacement field itself. For faster convergence, residual smoothing is performed, and the Jacobian computation is parallelized. The framework supports general constitutive laws as demonstrated by Neo-Hookean and Fung-type models.

Results: The inverse solver is verified for a variety of cases, including spatially varying elasticity and anisotropic fiber distributions, and its applicability to complex 3D geometries is demonstrated on a bioprosthetic heart valve. The residual-only formulation leads to faster optimization iterations compared to Physics-Informed Neural Network (PINN) approaches, which involve computationally expensive updates to a displacement-approximating network. Unlike the Neo-Hookean constitutive model, the Fung model, especially with element-wise recovery of parameters, exhibits nonuniqueness, where different spatial distributions of parameters can yield similar residual levels; in contrast, fiber direction recovery remains robust across all cases.

Conclusion: The inverse solver is capable of recovering nonlinear material properties and fiber directions even for complex 3D problems with large deformations such as heart valves.

Purpose: Collagen fibrils provide mechanical integrity to the extracellular matrix in a variety of biological tissues. In biomedical research, quantification of fibril diameters is essential to describe remodeling of the matrix that can occur during development, disease, and healing. However, current statistical methods to analyze differences in the distribution of fibril diameters have significant limitations. This study evaluated a rigorous alternative method, Statistical nonParametric Mapping (SnPM), to compare fibril diameter distributions.

Methods: Randomly generated simulated datasets and experimental datasets of fibril diameter distributions were analyzed using both conventional tests and SnPM. Results from each method were compared.

Results: Conventional statistical tests to compare fibril diameter distributions demonstrated limitations. Comparison of average diameters detected overall shifts in distributions but not more nuanced changes, while analysis of binned relative frequencies showed dependency on the choice of bin width and location. In contrast, comparison of kernel density estimated probability density functions using SnPM both detected differences between groups and located those differences to specific ranges of fibril diameters.

Conclusion: Comparative analysis of groups of fibril diameter distributions using SnPM overcomes the limitations of current techniques and provides a reliable and rigorous technique for these data. Other potential applications for SnPM in biomedical research abound, extending to other distributional or multidimensional data.

Purpose: Shoulder kinematic and muscular redundancy promotes considerable variability, obscuring possible insights into neuromuscular control and compensation mechanisms for muscle weakness or fatigue. The current study harnessed recent advancements in optimal control formulations for computational musculoskeletal models to determine potential neuromuscular control strategies to compensate for isolated muscle weakness.

Methods: A computational shoulder model characterized by independent clavicular, scapular, and humeral kinematics and 138 muscle elements was used. Optimal control-predicted thoracohumeral elevation kinematics were validated against published empirical kinematics. Force-generating capacity of the upper trapezius, middle trapezius, and lower trapezius, and serratus anterior were individually limited to 75%, 50%, and 25% maximal capacity to generate subsequent optimal control predictions of scapulothoracic kinematic changes associated with muscle weakness. Combined limited maximal force-generating capacity of lower trapezius and serratus anterior was also explored.

Results: Model-predicted scapulothoracic kinematics showed good agreement with reference data, yet some significant differences were identified below 55° thoracohumeral elevation. Fatigue-mediated kinematic changes were most apparent during sagittal plane elevation. Serratus anterior weakness displayed the largest scapulothoracic kinematic changes at all thresholds of limited force-generating capacity. It also prompted the largest compensatory muscle activity changes from other shoulder muscles, while upper trapezius weakness prompted very little compensatory changes in muscle activity.

Conclusion: Optimal control simulations were used to identify potential compensation mechanisms for shoulder muscle weakness and predict their effects on scapular kinematics. Findings suggest that thoracohumeral elevation in the scapular plane displayed less trapezius coactivity, both when 'weakened' and 'unweakened.' Thus, scapular plane tasks may isolate serratus anterior, while frontal plane tasks may achieve more balanced coactivation.

Purpose: Due to the small scale and complexity of the acinus, the demarcation between tissue and air is often unclear, making segmentation challenging. Accurate segmentation is essential for analyzing lung development, structural changes, and functional roles, offering insights into lung growth and disease. Traditional methods require large, annotated datasets, which are difficult to obtain, especially in medical imaging, where annotations are time-consuming and costly. Annotating mice lungs imaged by synchrotron radiation-based X-ray tomographic microscopy (SRXTM) is particularly difficult due to their small size, complexity, and subtle boundaries.

Methods: This paper presents a novel approach using transfer learning with limited data for segmenting mice lungs in SRXTM images. A U-net model, evaluated on a dataset of 59 image/mask pairs of lung sections, was pretrained on lung section images and then used to segment entire lungs in images. Segmentation accuracy improved through fine-tuning with transfer learning, using only 5 labeled entire lung image/mask pairs, with 2 pairs for testing. This highlights the potential of transfer learning in addressing the challenge of limited annotated data.

Results: The results show that U-net applied on the lung sections images achieves accurate prediction of the lung section masks, with the following metrics DSC = 0.9069, mIOU = 0.895, and BCE = 0.141. After transfer learning to entire lung images, segmentation results were DSC = 0.8915, mIOU = 0.8895, and BCE = 0.141.

Conclusion: Experimental results demonstrate the effectiveness of the proposed method, achieving competitive segmentation performance compared to traditional approaches, while requiring significantly fewer annotated images.

Background and objective: Computed tomography-based fractional flow reserve computation (CT-FFR) is widely used in clinical practice, based on its efficacy demonstrated in many studies. However, major assumptions remain with the outflow boundary conditions (BCs) representing coronary microvasculature, especially in hyperaemia. We here propose a novel method to estimate patient-specific microvascular response to hyperaemia for CT-FFR calculations, based on patients' routinely available demographic data.

Methods: A statistical model to predict microvascular flow response (MFR) from routinely collected patient demographic parameters was derived using PET-based perfusion data of 101 patients with coronary artery disease. CT-FFR computations were then conducted with patient-specific anatomical models and outflow BCs derived from various MFR models including the proposed approach. The FFR values were calculated for an independent test cohort of 10 patients who had undergone CT coronary angiography, CT perfusion imaging and invasive FFR measurement. Computed FFR values were compared against invasive FFR and other CT-FFR algorithms.

Results: A multivariate regression model predicting patient-specific MFR was derived as a function of sex, diabetes and smoking status of the patient. FFR values computed using our model agreed well with the invasive FFR (0.76 ± 0.09 vs. 0.75 ± 0.10, P = 0.217). The FFRs predicted with our model were also comparable to those calculated using outflow BC tuned with patient-specific perfusion data (FFR: 0.74 ± 0.10, P = 0.233 vs. invasive FFR) and showed marked improvement over the conventional approach (FFR: 0.68 ± 0.11, P = 0.004 vs. invasive FFR). Diagnostic accuracy vs. invasive FFR were 100, 91 and 82% for CT-FFR with CTP-based MFR, demography-based MFR, and conventional approach, respectively.

Discussion: The proposed demography-based MFR model significantly improves FFR computation accuracy compared with a typical conventional model that assumes constant, healthy and population average MFR. Although its diagnostic accuracy is slightly lower than that of CT-FFR calibrated with patient-specific perfusion imaging data (91 vs. 100%), the demography-based model offers a substantial practical advantage by not requiring additional non-standard data acquisition, such as perfusion imaging. Consequently, it shows strong potential as a practical enhancement to conventional CT-FFR algorithms.

Purpose: Established in silico frameworks for assessing Torsade de Pointes (TdP) risk primarily rely on single-cell electrophysiological biomarkers, which have demonstrated strong predictive capabilities. However, ventricular transmural electrophysiological heterogeneity is known to influence repolarization dynamics and arrhythmogenic mechanisms. Explicitly incorporating endocardium, epicardium, and mid-myocardium representations may enhance physiological interpretability, but direct integration of multi-cell features can introduce severe multicollinearity and compromise model stability. To address this challenge, we propose an ordinal logistic regression (OLR) framework that integrates multi-cell qNet information through probability averaging, preserving physiological context while maintaining robust statistical behavior.

Methods: $I{C}_{50}$ values and Hill coefficients for 28 CiPA drugs were implemented in two ventricular cell models, the CiPAORdV1.0 and the ORd in silico models. qNet was computed independently for endocardium, epicardium, and mid-myocardium cells. Cell-specific OLR models produced class probabilities that were then averaged to generate the final prediction. Performance was compared against single-cell and direct multi-cell implementations across Manual and ChanTest datasets.

Results: For CiPA-ORd v1.0 using the ChanTest dataset, qNet achieved substantial performance, with AUCs for ROC1 and ROC2 of 1.000 and 0.958, respectively, and also meeting seven "excellent" classification criteria. In the ORd model, probability averaging consistently improved performance for both the Manual and ChanTest datasets relative to single-cell and direct multi-cell approaches.

Conclusion: Probability-averaged integration of multi-cell qNet predictions mitigates multicollinearity while preserving physiological relevance, yielding more stable and accurate in silico TdP risk classification and supporting broader applicability to preclinical safety assessment.