Bioengineering

12 pages, 1033 KiB

Open AccessArticle

A Time-Series Approach for Machine Learning-Based Patient-Specific Quality Assurance of Radiosurgery Plans

by Simone Buzzi, Pietro Mancosu, Andrea Bresolin, Pasqualina Gallo, Francesco La Fauci, Francesca Lobefalo, Lucia Paganini, Marco Pelizzoli, Giacomo Reggiori, Ciro Franzese, Stefano Tomatis, Marta Scorsetti, Cristina Lenardi and Nicola Lambri

Bioengineering 2025, 12(8), 897; https://doi.org/10.3390/bioengineering12080897 - 21 Aug 2025

Viewed by 117

Abstract

Stereotactic radiosurgery (SRS) for multiple brain metastases can be delivered with a single isocenter and non-coplanar arcs, achieving highly conformal dose distributions at the cost of extreme modulation of treatment machine parameters. As a result, SRS plans are at a higher risk of [...] Read more.

Stereotactic radiosurgery (SRS) for multiple brain metastases can be delivered with a single isocenter and non-coplanar arcs, achieving highly conformal dose distributions at the cost of extreme modulation of treatment machine parameters. As a result, SRS plans are at a higher risk of patient-specific quality assurance (PSQA) failure compared to standard treatments. This study aimed to develop a machine-learning (ML) model to predict the PSQA outcome (gamma passing rate, GPR) of SRS plans. Five hundred and ninety-two consecutive patients treated between 2020 and 2024 were selected. GPR analyses were performed using a 3%/1 mm criterion and a 95% action limit for each arc. Fifteen plan complexity metrics were used as input features to predict the GPR of an arc. A stratified and a time-series approach were employed to split the data into training (1555 arcs), validation (389 arcs), and test (486 arcs) sets. The ML model achieved a mean absolute error of 2.6% on the test set, with a 0.83% median residual value (measured/predicted). Lower values of the measured GPR tended to be overestimated. Sensitivity and specificity were 93% and 56%, respectively. ML models for virtual QA of SRS can be integrated into clinical practice, facilitating more efficient PSQA approaches. Full article

(This article belongs to the Special Issue Radiation Imaging and Therapy for Biomedical Engineering)

► Show Figures

Figure 1

10 pages, 705 KiB

Open AccessArticle

Introducing Holographic Surgical Navigation in Pediatric Wilms’ Tumor Patients: A Feasibility Study During Total Nephrectomy

by Nick T. de Groot, Jasper M. van der Zee, Guus M. J. Bökkerink, Annemieke S. Littooij, Caroline C. C. Hulsker, Cecilia E. J. Terwisscha van Scheltinga, Cornelis P. van de Ven, Ruud C. Wortel, Aart J. Klijn, Marc H. W. A. Wijnen, Matthijs Fitski and Alida F. W. van der Steeg

Bioengineering 2025, 12(8), 896; https://doi.org/10.3390/bioengineering12080896 - 21 Aug 2025

Viewed by 98

Abstract

Wilms’ tumor is a common pediatric renal malignancy. In selected cases, nephron-sparing surgery (NSS) may be employed as part of the surgical approach. To prevent positive margins, optimal understanding of the tumor–kidney edge is essential. Augmented reality (AR) enables intraoperative visualization of patient-specific [...] Read more.

Wilms’ tumor is a common pediatric renal malignancy. In selected cases, nephron-sparing surgery (NSS) may be employed as part of the surgical approach. To prevent positive margins, optimal understanding of the tumor–kidney edge is essential. Augmented reality (AR) enables intraoperative visualization of patient-specific three-dimensional (3D) holograms. In this study, we aim to validate the clinical feasibility of a holographic landmark-based registration system in pediatric patients planned for a total nephrectomy (TN), to ensure that the holographic visualization will not influence surgical decision making. In a single-center prospective study, ten pediatric patients undergoing TN were included. Patient-specific 3D holograms were created from preoperative MRI, and intraoperatively landmark-based registration was performed using the HoloLens 2. Clinical feasibility was conducted using accuracy measurements, the System Usability Scale (SUS), and a self-developed questionnaire. Three out of ten patients had a successful registration with a median measured accuracy of 7.0 mm (Interquartile Range (IQR) 6–13.5) and a median SUS score of 75 (IQR 65–77.5). Surgeons reported improved depth perception and anatomical understanding. However, in seven out of ten patients, registration failed due to multiple reasons. The most important factors were large tumor volumes obstructing landmark placement and insufficient spatial distributions of the landmarks, causing rotational misalignment. Although AR showed potential in improving the depth perception and relation in anatomical structures, the landmark-based registration with the HoloLens 2 was currently deemed insufficient for clinical implementation in pediatric abdominal surgery. Full article

(This article belongs to the Special Issue Augmented Reality, Virtual Reality and Extended Reality Solutions for Biomedical Applications)

► Show Figures

Graphical abstract

18 pages, 1143 KiB

Open AccessArticle

Enhancing Clinical Decision Support with Adaptive Iterative Self-Query Retrieval for Retrieval-Augmented Large Language Models

by Srinivasagam Prabha, Cesar A. Gomez-Cabello, Syed Ali Haider, Ariana Genovese, Maissa Trabilsy, Nadia G. Wood, Sanjay Bagaria, Cui Tao and Antonio J. Forte

Bioengineering 2025, 12(8), 895; https://doi.org/10.3390/bioengineering12080895 - 21 Aug 2025

Viewed by 113

Abstract

Retrieval-Augmented Generation (RAG) offers a promising strategy to harness large language models (LLMs) for delivering up-to-date, accurate clinical guidance while reducing physicians’ cognitive burden, yet its effectiveness hinges on query clarity and structure. We propose an adaptive Self-Query Retrieval (SQR) framework that integrates [...] Read more.

Retrieval-Augmented Generation (RAG) offers a promising strategy to harness large language models (LLMs) for delivering up-to-date, accurate clinical guidance while reducing physicians’ cognitive burden, yet its effectiveness hinges on query clarity and structure. We propose an adaptive Self-Query Retrieval (SQR) framework that integrates three refinement modules—PICOT (Population, Intervention, Comparison, Outcome, Time), SPICE (Setting, Population, Intervention, Comparison, Evaluation), and Iterative Query Refinement (IQR)—to automatically restructure and iteratively enhance clinical questions until they meet predefined retrieval-quality thresholds. Implemented on Gemini-1.0 Pro, we benchmarked SQR using thirty postoperative rhinoplasty queries, evaluating responses for accuracy and relevance on a three-point Likert scale and for retrieval quality via precision, recall, and F1 score; statistical significance was assessed by one-way ANOVA with Tukey post-hoc testing. The full SQR pipeline achieved 87% accuracy (Likert 2.4 ± 0.7) and 100% relevance (Likert 3.0 ± 0.0), significantly outperforming a non-refined RAG baseline (50% accuracy, 80% relevance; p < 0.01 and p = 0.03). Precision, recall, and F1 rose from 0.17, 0.39 and 0.24 to 0.53, 1.00, and 0.70, respectively, while PICOT-only and SPICE-only variants yielded intermediate improvements. These findings demonstrate that automated structuring and iterative enhancement of queries via SQR substantially elevate LLM-based clinical decision support, and its model-agnostic architecture enables rapid adaptation across specialties, data sources, and LLM platforms. Full article

(This article belongs to the Special Issue Application of Artificial Intelligence in Complex Diseases)

► Show Figures

Figure 1

17 pages, 5300 KiB

Open AccessArticle

Multimodal Integration Enhances Tissue Image Information Content: A Deep Feature Perspective

by Fatemehzahra Darzi and Thomas Bocklitz

Bioengineering 2025, 12(8), 894; https://doi.org/10.3390/bioengineering12080894 - 21 Aug 2025

Viewed by 132

Abstract

Multimodal imaging techniques have the potential to enhance the interpretation of histology by offering additional molecular and structural information beyond that accessible through hematoxylin and eosin (H&E) staining alone. Here, we present a quantitative approach for comparing the information content of different image [...] Read more.

Multimodal imaging techniques have the potential to enhance the interpretation of histology by offering additional molecular and structural information beyond that accessible through hematoxylin and eosin (H&E) staining alone. Here, we present a quantitative approach for comparing the information content of different image modalities, such as H&E and multimodal imaging. We used a combination of deep learning and radiomics-based feature extraction with different information markers, implemented in Python 3.12, to compare the information content of the H&E stain, multimodal imaging, and the combined dataset. We also compared the information content of individual channels in the multimodal image and of different Coherent Anti-Stokes Raman Scattering (CARS) microscopy spectral channels. The quantitative measurements of information that we utilized were Shannon entropy, inverse area under the curve (1-AUC), the number of principal components describing 95% of the variance (PC95), and inverse power law fitting. For example, the combined dataset achieved an entropy value of 0.5740, compared to 0.5310 for H&E and 0.5385 for the multimodal dataset using MobileNetV2 features. The number of principal components required to explain 95 percent of the variance was also highest for the combined dataset, with 62 components, compared to 33 for H&E and 47 for the multimodal dataset. These measurements consistently showed that the combined datasets provide more information. These observations highlight the potential of multimodal combinations to enhance image-based analyses and provide a reproducible framework for comparing imaging approaches in digital pathology and biomedical image analysis. Full article

(This article belongs to the Special Issue Medical Imaging Analysis: Current and Future Trends)

► Show Figures

Figure 1

28 pages, 814 KiB

Open AccessReview

Functional Carbon-Based Materials for Blood Purification: Recent Advances Toward Improved Treatment of Renal Failure and Patient Quality of Life

by Abolfazl Mozaffari, Farbod Alimohammadi and Mazeyar Parvinzadeh Gashti

Bioengineering 2025, 12(8), 893; https://doi.org/10.3390/bioengineering12080893 - 21 Aug 2025

Viewed by 175

Abstract

The accumulation of blood toxins, including urea, uric acid, creatinine, bilirubin, p-cresyl sulfate, and indoxyl sulfate, poses severe health risks for patients with renal failure. Effective removal strategies are essential to mitigate complications associated with chronic kidney disease (CKD) and improve patient outcomes. [...] Read more.

The accumulation of blood toxins, including urea, uric acid, creatinine, bilirubin, p-cresyl sulfate, and indoxyl sulfate, poses severe health risks for patients with renal failure. Effective removal strategies are essential to mitigate complications associated with chronic kidney disease (CKD) and improve patient outcomes. Functional carbon-based materials, such as activated carbon (activated charcoal) and graphene oxide, have emerged as promising adsorbents due to their large surface area, adjustable porosity, and biocompatibility. This review comprehensively explores the latest advancements in carbon-based materials for blood purification across three key therapeutic modalities: (1) Hemoperfusion, where activated and modified carbonaceous materials enhance the adsorption of small-molecule and protein-bound toxins; (2) Hemodialysis, where functionalized carbon materials improve clearance rates and reduce treatment duration; and (3) Oral Therapeutics, where orally administered carbon adsorbents show potential in lowering systemic toxin levels in CKD patients. Furthermore, we present a comparative analysis of these approaches, highlighting their advantages, limitations, and future research directions for optimizing carbon-based detoxification strategies. The findings discussed in this review emphasize the significance of material engineering in advancing blood purification technologies. By enhancing the efficiency of toxin removal, carbon-based materials have the potential to revolutionize renal failure treatment, offering improved clinical outcomes and enhanced patient quality of life. Full article

(This article belongs to the Special Issue Emerging Nanomaterials and Nanotechnologies for Biomedical and Related Applications)

► Show Figures

Figure 1

12 pages, 813 KiB

Open AccessArticle

Evaluating SnapshotNIR for Tissue Oxygenation Measurement Across Skin Types After Mastectomy

by Saif Badran, Sara Saffari, William R. Moritz, Gary B. Skolnick, Amanda M. Westman, Mitchell A. Pet and Justin M. Sacks

Bioengineering 2025, 12(8), 892; https://doi.org/10.3390/bioengineering12080892 - 21 Aug 2025

Viewed by 126

Abstract

Accurate monitoring of mastectomy skin flap (MSF) perfusion is critical, especially in patients with darker skin pigmentation at higher risk of misdiagnosed tissue ischemia. Near-infrared spectroscopy (NIRS) devices, such as SnapshotNIR, offer real-time tissue oxygen saturation measurements (StO₂), but their accuracy [...] Read more.

Accurate monitoring of mastectomy skin flap (MSF) perfusion is critical, especially in patients with darker skin pigmentation at higher risk of misdiagnosed tissue ischemia. Near-infrared spectroscopy (NIRS) devices, such as SnapshotNIR, offer real-time tissue oxygen saturation measurements (StO₂), but their accuracy across skin pigmentation levels remains unexplored. This quasi-experimental study included 33 patients undergoing mastectomy. MSF edge ΔStO₂, defined as preoperative minus postoperative StO_2, was measured using SnapshotNIR device (Kent Imaging, Calgary, AB, Canada) pre- and post-mastectomy. By definition, a positive ΔStO₂ indicates a decrease in tissue oxygenation, while a negative ΔStO₂ indicates an increase relative to baseline. ΔStO₂ was analyzed against Fitzpatrick scores to assess skin pigmentation impact on measurement accuracy. ΔStO₂ (mean ± SD) progressively decreased with increasing Fitzpatrick score: 14.0 ± 22.98 for score 1, 6.87 ± 17.45 for score 2, −3.13 ± 6.89 for score 3, and −40.75 ± 22.27 for score 5, indicating a shift from positive to negative O₂ change. Fitzpatrick scores significantly correlated with ΔStO₂ (ρ = −0.392, p = 0.016). ANOVA confirmed differences (p = 0.008), with Tukey’s post hoc testing showing significant differences between Fitzpatrick scores 1 and 5 (p = 0.022), and 2 and 5 (p = 0.006). SnapshotNIR technology demonstrated measurable sensitivity for detecting changes in StO₂ and predicting ischemia; however, NIRS-based devices may overestimate oxygenation in darker skin pigmentation, highlighting a need for device calibration to improve accuracy across skin tones. Full article

(This article belongs to the Special Issue Biomedical Engineering and Biomarkers in Plastic and Reconstructive Surgery Applications)

► Show Figures

Graphical abstract

15 pages, 5996 KiB

Open AccessArticle

A High-Fidelity mmWave Radar Dataset for Privacy-Sensitive Human Pose Estimation

by Yuanzhi Su, Huiying (Cynthia) Hou, Haifeng Lan and Christina Zong-Hao Ma

Bioengineering 2025, 12(8), 891; https://doi.org/10.3390/bioengineering12080891 - 21 Aug 2025

Viewed by 109

Abstract

Human pose estimation (HPE) in privacy-sensitive environments such as healthcare facilities and smart homes demands non-visual sensing solutions. Millimeter-wave (mmWave) radar emerges as a promising alternative, yet its development is hindered by the scarcity of high-fidelity datasets with accurate annotations. This paper introduces [...] Read more.

Human pose estimation (HPE) in privacy-sensitive environments such as healthcare facilities and smart homes demands non-visual sensing solutions. Millimeter-wave (mmWave) radar emerges as a promising alternative, yet its development is hindered by the scarcity of high-fidelity datasets with accurate annotations. This paper introduces mmFree-Pose, the first dedicated mmWave radar dataset specifically designed for privacy-preserving HPE. Collected through a novel visual-free framework that synchronizes mmWave radar with VDSuit-Full motion-capture sensors, our dataset covers 10+ actions, from basic gestures to complex falls. Each sample provides (i) raw 3D point clouds with Doppler velocity and intensity, (ii) precise 23-joint skeletal annotations, and (iii) full-body motion sequences in privacy-critical scenarios. Crucially, all data is captured without the use of visual sensors, ensuring fundamental privacy protection by design. Unlike conventional approaches that rely on RGB or depth cameras, our framework eliminates the risk of visual data leakage while maintaining high annotation fidelity. The dataset also incorporates scenarios involving occlusions, different viewing angles, and multiple subject variations to enhance generalization in real-world applications. By providing a high-quality and privacy-compliant dataset, mmFree-Pose bridges the gap between RF sensing and home monitoring applications, where safeguarding personal identity and behavior remains a critical concern. Full article

(This article belongs to the Special Issue Biomechanics and Motion Analysis)

► Show Figures

Figure 1

22 pages, 5543 KiB

Open AccessArticle

Mapping Emerging Scientific Trends in Chronic Skin Disorders Using Machine Learning-Based Bibliometrics

by Nicoleta Cirstea, Andrei-Flavius Radu, Delia Mirela Tit, Ada Radu, Gabriela S. Bungau, Laura Maria Endres and Paul Andrei Negru

Bioengineering 2025, 12(8), 890; https://doi.org/10.3390/bioengineering12080890 - 20 Aug 2025

Viewed by 188

Abstract

Chronic dermatologic diseases are characterized by pathophysiologic complexity and the existence of many unmet patient management needs that can contribute to treatment failure, with poor adherence being a major issue. This study aims to identify key topics in this field, using the Web [...] Read more.

Chronic dermatologic diseases are characterized by pathophysiologic complexity and the existence of many unmet patient management needs that can contribute to treatment failure, with poor adherence being a major issue. This study aims to identify key topics in this field, using the Web of Science database. To perform this analysis, tools such as VOSviewer, Bibliometrix, and Excel were used. A Python script leveraging machine learning algorithms was developed to standardize terminology. The initial search yielded 35,373 documents, which were then refined to 12,952 publications spanning 1975 to 2024 through parameter optimization. The study found an increasing interest in this research domain, with a notable surge in 2019. The analysis identified the United States, Germany, and England as the most prolific countries in terms of scientific output. Canada ranked sixth in total document production, but its documents received the highest average citations, reflecting a significant impact. Normalization analysis revealed Italy as the most specialized country in chronic skin disease research relative to total national research output. Trend analysis revealed an evolution in research topics, particularly after 2020, with a growing focus on personalized treatment methods and long-term treatment outcomes. The study highlighted international collaboration, especially among countries with cultural or regional connections, such as those within the European Union. It underscores the growing need for continuous updates and the increasing global focus on chronic skin diseases, highlighting the critical role of staying current with emerging trends to drive advancements in treatment and patient care. Full article

(This article belongs to the Section Biosignal Processing)

► Show Figures

Figure 1

13 pages, 941 KiB

Open AccessArticle

Biomechanical Characterisation of Gait in Older Adults: A Cross-Sectional Study Using Inertial Sensor-Based Motion Capture

by Anna Letournel, Madalena Marques, Ricardo Vigário, Carla Quintão and Cláudia Quaresma

Bioengineering 2025, 12(8), 889; https://doi.org/10.3390/bioengineering12080889 - 20 Aug 2025

Viewed by 189

Abstract

The ageing of the global population, especially in developed countries, is driving significant societal changes. In Portugal, demographic data reflect a marked increase in the ageing index. Understanding gait alterations associated with ageing is essential for the early detection of mobility decline and [...] Read more.

The ageing of the global population, especially in developed countries, is driving significant societal changes. In Portugal, demographic data reflect a marked increase in the ageing index. Understanding gait alterations associated with ageing is essential for the early detection of mobility decline and fall risk. This study aimed to analyse gait patterns in older adults to contribute to a biomechanical ageing profile. Thirty-six community-dwelling older adults (29 female, 7 male; mean age: 74 years) participated. Gait data were collected using the Xsens full-body motion capture system, which combines inertial sensors with biomechanical modelling and sensor fusion. Spatiotemporal and kinematic parameters were analysed using descriptive statistics. Compared to younger adult norms, participants showed increased stance and double support phases, reduced swing phase, and lower gait speed, stride length, and cadence, with greater step width. Kinematic data showed reduced peak plantar flexion, knee flexion, and hip extension, but increased dorsiflexion peaks—adaptations aimed at stability. Despite a limited sample size and lack of clinical subgroups, results align with age-related gait literature. Findings support the utility of wearable systems like Xsens in capturing clinically relevant gait changes, contributing to normative biomechanical profiling and future mobility interventions. Full article

(This article belongs to the Section Biosignal Processing)

► Show Figures

Graphical abstract

33 pages, 7581 KiB

Open AccessArticle

Effect of Bone Quality, Implant Length, and Loading Timing on Stress Transmission in the Posterior Mandible: A Finite Element Analysis

by Ladise Ceylin Has and Recep Orbak

Bioengineering 2025, 12(8), 888; https://doi.org/10.3390/bioengineering12080888 - 20 Aug 2025

Viewed by 137

Abstract

This study aimed to evaluate the biomechanical effects of implant length, mandibular morphology, graft application, loading timing, and force direction on peri-implant stress distribution using finite element analysis (FEA). Five mandibular models representing normal, atrophic, and graft-augmented conditions were constructed. Each model was [...] Read more.

This study aimed to evaluate the biomechanical effects of implant length, mandibular morphology, graft application, loading timing, and force direction on peri-implant stress distribution using finite element analysis (FEA). Five mandibular models representing normal, atrophic, and graft-augmented conditions were constructed. Each model was analyzed with 6 mm and 12 mm Straumann Standard implants under two loading types, vertical (200 N) and oblique (100 N at 30°), across three loading protocols (immediate, early, and delayed). Stress analysis was conducted using von Mises and principal stress criteria, focusing on cortical and trabecular bone, the implant–abutment complex, and the mandibular canal. Under vertical loading, increasing the implant length from 6 mm to 12 mm reduced the maximum tensile stresses in trabecular bone from 0.930 MPa to 0.475 MPa (an approximate 49% decrease). However, oblique loading caused a substantial increase in stresses in all regions, with trabecular compressive stress reaching up to −19.102 MPa and cortical tensile stress up to 179.798 MPa in the atrophic mandible. Graft application significantly reduced peri-implant stresses; for example, maximum compressive stress in the cortical bone decreased from −227.051 MPa in the atrophic model to −13.395 MPa in the grafted model under similar loading conditions. Although the graft donor site was not explicitly modeled, the graft material (Bio-Oss) was anatomically positioned in the posterior mandible to simulate buccolingual augmentation and its biomechanical effects. Stress concentrations around the mandibular canal remained below the 6 MPa threshold for neurovascular injury in all scenarios, indicating a biomechanically safe outcome. These findings indicate that oblique loading and reduced bone volume may compromise implant survival, whereas graft application plays a critical role in mitigating stress levels and enhancing biomechanical stability. The study also emphasizes the importance of considering force direction and bone quality in clinical planning, and highlights the novelty of combining graft simulation with FEA to assess its protective role beyond implant length alone. Full article

(This article belongs to the Section Biomechanics and Sports Medicine)

► Show Figures

Figure 1

42 pages, 2529 KiB

Open AccessReview

Artificial Intelligence in Sports Biomechanics: A Scoping Review on Wearable Technology, Motion Analysis, and Injury Prevention

by Marouen Souaifi, Wissem Dhahbi, Nidhal Jebabli, Halil İbrahim Ceylan, Manar Boujabli, Raul Ioan Muntean and Ismail Dergaa

Bioengineering 2025, 12(8), 887; https://doi.org/10.3390/bioengineering12080887 - 20 Aug 2025

Viewed by 533

Abstract

Aim: This scoping review examines the application of artificial intelligence (AI) in sports biomechanics, with a focus on enhancing performance and preventing injuries. The review addresses key research questions, including primary AI methods, their effectiveness in improving athletic performance, their potential for injury [...] Read more.

Aim: This scoping review examines the application of artificial intelligence (AI) in sports biomechanics, with a focus on enhancing performance and preventing injuries. The review addresses key research questions, including primary AI methods, their effectiveness in improving athletic performance, their potential for injury prediction, sport-specific applications, strategies for translating knowledge, ethical considerations, and remaining research gaps. Following the PRISMA-ScR guidelines, a comprehensive literature search was conducted across five databases (PubMed/MEDLINE, Web of Science, IEEE Xplore, Scopus, and SPORTDiscus), encompassing studies published between January 2015 and December 2024. After screening 3248 articles, 73 studies met the inclusion criteria (Cohen’s kappa = 0.84). Data were collected on AI techniques, biomechanical parameters, performance metrics, and implementation details. Results revealed a shift from traditional statistical models to advanced machine learning methods. Based on moderate-quality evidence from 12 studies, convolutional neural networks reached 94% agreement with international experts in technique assessment. Computer vision demonstrated accuracy within 15 mm compared to marker-based systems (6 studies, moderate quality). AI-driven training plans showed 25% accuracy improvements (4 studies, limited evidence). Random forest models predicted hamstring injuries with 85% accuracy (3 studies, moderate quality). Learning management systems enhanced knowledge transfer, raising coaches’ understanding by 45% and athlete adherence by 3.4 times. Implementing integrated AI systems resulted in a 23% reduction in reinjury rates. However, significant challenges remain, including standardizing data, improving model interpretability, validating models in real-world settings, and integrating them into coaching routines. In summary, incorporating AI into sports biomechanics marks a groundbreaking advancement, providing analytical capabilities that surpass traditional techniques. Future research should focus on creating explainable AI, applying rigorous validation methods, handling data ethically, and ensuring equitable access to promote the widespread and responsible use of AI across all levels of competitive sports. Full article

(This article belongs to the Section Biomechanics and Sports Medicine)

► Show Figures

Figure 1

23 pages, 1615 KiB

Open AccessReview

Current Mechanobiological Pathways and Therapies Driving Spinal Health

by Rahul Kumar, Kyle Sporn, Harlene Kaur, Akshay Khanna, Phani Paladugu, Nasif Zaman and Alireza Tavakkoli

Bioengineering 2025, 12(8), 886; https://doi.org/10.3390/bioengineering12080886 - 20 Aug 2025

Viewed by 273

Abstract

Spinal health depends on the dynamic interplay between mechanical forces, biochemical signaling, and cellular behavior. This review explores how key molecular pathways, including integrin, yeas-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), Piezo, and Wingless/Integrated (Wnt) with β-catenin, actively shape the [...] Read more.

Spinal health depends on the dynamic interplay between mechanical forces, biochemical signaling, and cellular behavior. This review explores how key molecular pathways, including integrin, yeas-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), Piezo, and Wingless/Integrated (Wnt) with β-catenin, actively shape the structural and functional integrity of spinal tissues. These signaling mechanisms respond to physical cues and interact with inflammatory mediators such as interleukin-1 beta (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor alpha (TNF-α), driving changes that lead to disc degeneration, vertebral fractures, spinal cord injury, and ligament failure. New research is emerging that shows scaffold designs that can directly harness these pathways. Further, new stem cell-based therapies have been shown to promote disc regeneration through targeted differentiation and paracrine signaling. Interestingly, many novel bone and ligament scaffolds are modulating anti-inflammatory signals to enhance tissue repair and integration, as well as prevent scaffold degradation. Neural scaffolds are also arising. These mimic spinal biomechanics and activate Piezo signaling to guide axonal growth and restore motor function. Scientists have begun combining these biological platforms with brain–computer interface technology to restore movement and sensory feedback in patients with severe spinal damage. Although this technology is not fully clinically ready, this field is advancing rapidly. As implantable technology can now mimic physiological processes, molecular signaling, biomechanical design, and neurotechnology opens new possibilities for restoring spinal function and improving the quality of life for individuals with spinal disorders. Full article

(This article belongs to the Special Issue Biomechanics and Mechanobiology in Cell and Tissue Engineering)

► Show Figures

Figure 1

10 pages, 511 KiB

Open AccessArticle

Improving Benign and Malignant Classifications in Mammography with ROI-Stratified Deep Learning

by Kenji Yoshitsugu, Kazumasa Kishimoto and Tadamasa Takemura

Bioengineering 2025, 12(8), 885; https://doi.org/10.3390/bioengineering12080885 - 20 Aug 2025

Viewed by 192

Abstract

Deep learning has achieved widespread adoption for medical image diagnosis, with extensive research dedicated to mammographic image analysis for breast cancer screening. This study investigates the hypothesis that incorporating region-of-interest (ROI) mask information for individual mammographic images during deep learning can improve the [...] Read more.

Deep learning has achieved widespread adoption for medical image diagnosis, with extensive research dedicated to mammographic image analysis for breast cancer screening. This study investigates the hypothesis that incorporating region-of-interest (ROI) mask information for individual mammographic images during deep learning can improve the accuracy of benign/malignant diagnoses. Swin Transformer and ConvNeXtV2 deep learning models were used to evaluate their performance on the public VinDr and CDD-CESM datasets. Our approach involved stratifying mammographic images based on the presence or absence of ROI masks, performing independent training and prediction for each subgroup, and subsequently merging the results. Baseline prediction metrics (sensitivity, specificity, F-score, and accuracy) without ROI-stratified separation were the following: VinDr/Swin Transformer (0.00, 1.00, 0.00, 0.85), VinDr/ConvNeXtV2 (0.00, 1.00, 0.00, 0.85), CDD-CESM/Swin Transformer (0.29, 0.68, 0.41, 0.48), and CDD-CESM/ConvNeXtV2 (0.65, 0.65, 0.65, 0.65). Subsequent analysis with ROI-stratified separation demonstrated marked improvements in these metrics: VinDr/Swin Transformer (0.93, 0.87, 0.90, 0.87), VinDr/ConvNeXtV2 (0.90, 0.86, 0.88, 0.87), CDD-CESM/Swin Transformer (0.65, 0.65, 0.65, 0.65), and CDD-CESM/ConvNeXtV2 (0.74, 0.61, 0.67, 0.68). These findings provide compelling evidence that validate our hypothesis and affirm the utility of considering ROI mask information for enhanced diagnostic accuracy in mammography. Full article

(This article belongs to the Special Issue Artificial Intelligence Revolution in Biomedical Image and Signal Processing: Innovations and Applications)

► Show Figures

Figure 1

18 pages, 6550 KiB

Open AccessArticle

scOTM: A Deep Learning Framework for Predicting Single-Cell Perturbation Responses with Large Language Models

by Yuchen Wang, Tianchi Lu, Xingjian Chen, Zhongyu Yao and Ka-Chun Wong

Bioengineering 2025, 12(8), 884; https://doi.org/10.3390/bioengineering12080884 - 20 Aug 2025

Viewed by 263

Abstract

Modeling drug-induced transcriptional responses at the single-cell level is essential for advancing human healthcare, particularly in understanding disease mechanisms, assessing therapeutic efficacy, and anticipating adverse effects. However, existing approaches often impose a rigid constraint by enforcing pointwise alignment of latent representations to a [...] Read more.

Modeling drug-induced transcriptional responses at the single-cell level is essential for advancing human healthcare, particularly in understanding disease mechanisms, assessing therapeutic efficacy, and anticipating adverse effects. However, existing approaches often impose a rigid constraint by enforcing pointwise alignment of latent representations to a standard normal prior, which limits expressiveness and results in biologically uninformative embeddings, especially in complex biological systems. Additionally, many methods inadequately address the challenges of unpaired data, typically relying on naive averaging strategies that ignore cell-type specificity and intercellular heterogeneity. To overcome these limitations, we propose scOTM, a deep learning framework designed to predict single-cell perturbation responses from unpaired data, focusing on generalization to unseen cell types. scOTM integrates prior biological knowledge of perturbations and cellular states, derived from large language models specialized for molecular and single-cell corpora. These informative representations are incorporated into a variational autoencoder with maximum mean discrepancy regularization, allowing flexible modeling of transcriptional shifts without imposing a strict constraint of alignment to a standard normal prior. scOTM further employs optimal transport to establish an efficient and interpretable mapping between control and perturbed distributions, effectively capturing the transcriptional shifts underlying response variation. Extensive experiments demonstrate that scOTM outperforms existing methods in predicting whole-transcriptome responses and identifying top differentially expressed genes. Furthermore, scOTM exhibits superior robustness in data-limited settings and strong generalization capabilities across cell types. Full article

(This article belongs to the Section Biosignal Processing)

► Show Figures

Graphical abstract

14 pages, 1955 KiB

Open AccessArticle

Dynamic Behavior of the Stenting & Shielding Hernia System Fosters Neomyogenesis in Experimental Porcine Model

by Giuseppe Amato, Roberto Puleio, Antonino Agrusa, Vito Rodolico, Luca Cicero, Giovanni Cassata, Giuseppe Di Buono, Emanuele Battaglia, Claudia Neto, Giorgio Romano, William Ra and Giorgio Romano

Bioengineering 2025, 12(8), 883; https://doi.org/10.3390/bioengineering12080883 - 19 Aug 2025

Viewed by 190

Abstract

Despite significant advancements, prosthetic hernia repair continues to face unacceptably high complication rates. These likely stem from poor biological responses, such as stiff scar tissue leading to mesh shrinkage. To overcome these issues, the Stenting and Shielding (S&S) Hernia System, a newly designed [...] Read more.

Despite significant advancements, prosthetic hernia repair continues to face unacceptably high complication rates. These likely stem from poor biological responses, such as stiff scar tissue leading to mesh shrinkage. To overcome these issues, the Stenting and Shielding (S&S) Hernia System, a newly designed 3D dynamic device, has been developed for dissection-free laparoscopic placement to permanently obliterate hernia defects. Unlike conventional meshes, this device induces a regenerative biological response, promoting viable tissue growth rather than fibrotic plaque formation. In a porcine experimental model, the S&S device demonstrated the development of a great amount of muscle fibers, alongside nervous and vascular structures, within well-perfused connective tissue. Histological analysis of biopsy specimens excised from the experimental animals revealed progressive muscle fiber maturation from early myocyte development in the short term to fully developed muscle bundles in the long term. The enhanced biological response observed with the S&S device suggests a promising shift in hernia repair, potentially reversing the degenerative processes of hernia formation and promoting tissue regeneration. The S&S Hernia System described here can be classified not merely as a conventional hernia implant, but as part of a new category of hernia devices: the dynamic regenerative scaffold. Full article

(This article belongs to the Section Nanobiotechnology and Biofabrication)

► Show Figures

Figure 1

24 pages, 1087 KiB

Open AccessArticle

Supervised Learning and Large Language Model Benchmarks on Mental Health Datasets: Cognitive Distortions and Suicidal Risks in Chinese Social Media

by Hongzhi Qi, Guanghui Fu, Jianqiang Li, Changwei Song, Wei Zhai, Dan Luo, Shuo Liu, Yijing Yu, Bingxiang Yang and Qing Zhao

Bioengineering 2025, 12(8), 882; https://doi.org/10.3390/bioengineering12080882 - 19 Aug 2025

Viewed by 379

Abstract

On social media, users often express their personal feelings, which may exhibit cognitive distortions or even suicidal tendencies on certain specific topics. Early recognition of these signs is critical for effective psychological intervention. In this paper, we introduce two novel datasets from Chinese [...] Read more.

On social media, users often express their personal feelings, which may exhibit cognitive distortions or even suicidal tendencies on certain specific topics. Early recognition of these signs is critical for effective psychological intervention. In this paper, we introduce two novel datasets from Chinese social media: SOS-HL-1K for suicidal risk classification, which contains 1249 posts, and SocialCD-3K, a multi-label classification dataset for cognitive distortion detection that contains 3407 posts. We conduct a comprehensive evaluation using two supervised learning methods and eight large language models (LLMs) on the proposed datasets. From the prompt engineering perspective, we experiment with two types of prompt strategies, including four zero-shot and five few-shot strategies. We also evaluate the performance of the LLMs after fine-tuning on the proposed tasks. Experimental results show a significant performance gap between prompted LLMs and supervised learning. Our best supervised model achieves strong results, with an F1-score of 82.76% for the high-risk class in the suicide task and a micro-averaged F1-score of 76.10% for the cognitive distortion task. Without fine-tuning, the best-performing LLM lags by 6.95 percentage points in the suicide task and a more pronounced 31.53 points in the cognitive distortion task. Fine-tuning substantially narrows this performance gap to 4.31% and 3.14% for the respective tasks. While this research highlights the potential of LLMs in psychological contexts, it also shows that supervised learning remains necessary for more challenging tasks. Full article

(This article belongs to the Section Biosignal Processing)

► Show Figures

Figure 1

14 pages, 3334 KiB

Open AccessArticle

Development of a Computationally Efficient CFD Method for Blood Flow Analysis Following Flow Diverter Stent Deployment and Its Application to Treatment Planning

by Soichiro Fujimura, Haruki Kanebayashi, Kostadin Karagiozov, Tohru Sano, Shunsuke Hataoka, Michiyasu Fuga, Issei Kan, Hiroyuki Takao, Toshihiro Ishibashi, Makoto Yamamoto and Yuichi Murayama

Bioengineering 2025, 12(8), 881; https://doi.org/10.3390/bioengineering12080881 - 19 Aug 2025

Viewed by 244

Abstract

Intracranial aneurysms are a serious cerebrovascular condition with a risk of subarachnoid hemorrhage due to rupture, leading to high mortality and morbidity. Flow Diverter Stents (FDSs) have become an important endovascular treatment option for unruptured large or wide-neck aneurysms. Hemodynamic factors significantly influence [...] Read more.

Intracranial aneurysms are a serious cerebrovascular condition with a risk of subarachnoid hemorrhage due to rupture, leading to high mortality and morbidity. Flow Diverter Stents (FDSs) have become an important endovascular treatment option for unruptured large or wide-neck aneurysms. Hemodynamic factors significantly influence treatment outcomes in aneurysms treated with FDSs, and Computational Fluid Dynamics (CFD) has been widely used to evaluate post-deployment flow characteristics. However, conventional wire-resolved CFD methods require extremely fine meshes to reconstruct individual FDS wires, resulting in prohibitively high computational costs. This severely limits their feasibility for use in clinical treatment planning, where fast and robust simulations are essential. To address this limitation, we developed a computationally efficient CFD method that incorporates a porous media model accounting for local variations in wire density after FDS deployment. Based on Virtual Stent Simulation, the FDS region was defined as a hollow cylindrical domain with spatially varying resistance derived from cell-specific wire density. We validated the proposed method using 15 clinical cases, demonstrating close agreement with conventional wire-resolved CFD results. Relative errors in key hemodynamic parameters, including velocity, shear rate, inflow rate, and turnover time, were within 5%, with correlation coefficients exceeding 0.98. The number of grid elements, the data size, and total analysis time were reduced by over 90%. The method also allowed comparison between Total-Filling (OKM Grade A) and Occlusion (Grade D) cases, and evaluation of different FDS sizing, positioning, and coil-assisted strategies. The proposed method enables practical and efficient CFD analysis following FDS treatment and supports hemodynamics-based treatment planning of aneurysms. Full article

(This article belongs to the Section Biomedical Engineering and Biomaterials)

► Show Figures

Figure 1

10 pages, 941 KiB

Open AccessArticle

Therapeutic Role of Functional Massage in Attenuating Exercise-Induced Neuromuscular Fatigue

by Zahraa Darwich, Alaa Issa, Emma Parkin, Jada Young, Marie Eve Pepin and Moh H. Malek

Bioengineering 2025, 12(8), 880; https://doi.org/10.3390/bioengineering12080880 - 16 Aug 2025

Viewed by 417

Abstract

Background: Functional massage is a soft tissue intervention that combines tissue compression with specific joint movements to enhance muscle function, improve joint mobility and reduce pain. The physical working capacity at the fatigue threshold (PWC_FT) uses surface electromyography to determine the [...] Read more.

Background: Functional massage is a soft tissue intervention that combines tissue compression with specific joint movements to enhance muscle function, improve joint mobility and reduce pain. The physical working capacity at the fatigue threshold (PWC_FT) uses surface electromyography to determine the highest exercise intensity that can be sustained indefinitely. The purpose of this study, therefore, was to examine the influence of FM on a multi-joint exercise such as cycle ergometry. Methods: Twelve healthy college-aged men volunteered for the current study. On two occasions, separated by seven days and in randomized order, subjects either completed a 14 min FM on both legs prior to an incremental cycle ergometer test to determine PWC_FT, or rested for 14 min before performing the same cycling test. Results: The paired samples t-tests revealed a significant (p < 0.05) difference for the absolute and relative PWC_FT values between the no-FM and FM conditions. Conclusions: These results indicate that FM may delay the onset of neuromuscular fatigue for whole-body exercise. Full article

(This article belongs to the Special Issue Physical Therapy and Rehabilitation)

► Show Figures

Figure 1

18 pages, 1752 KiB

Open AccessSystematic Review

Beyond Post hoc Explanations: A Comprehensive Framework for Accountable AI in Medical Imaging Through Transparency, Interpretability, and Explainability

by Yashbir Singh, Quincy A. Hathaway, Varekan Keishing, Sara Salehi, Yujia Wei, Natally Horvat, Diana V. Vera-Garcia, Ashok Choudhary, Almurtadha Mula Kh, Emilio Quaia and Jesper B Andersen

Bioengineering 2025, 12(8), 879; https://doi.org/10.3390/bioengineering12080879 - 15 Aug 2025

Viewed by 724

Abstract

The integration of artificial intelligence (AI) in medical imaging has revolutionized diagnostic capabilities, yet the black-box nature of deep learning models poses significant challenges for clinical adoption. Current explainable AI (XAI) approaches, including SHAP, LIME, and Grad-CAM, predominantly focus on post hoc explanations [...] Read more.

The integration of artificial intelligence (AI) in medical imaging has revolutionized diagnostic capabilities, yet the black-box nature of deep learning models poses significant challenges for clinical adoption. Current explainable AI (XAI) approaches, including SHAP, LIME, and Grad-CAM, predominantly focus on post hoc explanations that may inadvertently undermine clinical decision-making by providing misleading confidence in AI outputs. This paper presents a systematic review and meta-analysis of 67 studies (covering 23 radiology, 19 pathology, and 25 ophthalmology applications) evaluating XAI fidelity, stability, and performance trade-offs across medical imaging modalities. Our meta-analysis of 847 initially identified studies reveals that LIME achieves superior fidelity (0.81, 95% CI: 0.78–0.84) compared to SHAP (0.38, 95% CI: 0.35–0.41) and Grad-CAM (0.54, 95% CI: 0.51–0.57) across all modalities. Post hoc explanations demonstrated poor stability under noise perturbation, with SHAP showing 53% degradation in ophthalmology applications (ρ = 0.42 at 10% noise) compared to 11% in radiology (ρ = 0.89). We demonstrate a consistent 5–7% AUC performance penalty for interpretable models but identify modality-specific stability patterns suggesting that tailored XAI approaches are necessary. Based on these empirical findings, we propose a comprehensive three-pillar accountability framework that prioritizes transparency in model development, interpretability in architecture design, and a cautious deployment of post hoc explanations with explicit uncertainty quantification. This approach offers a pathway toward genuinely accountable AI systems that enhance rather than compromise clinical decision-making quality and patient safety. Full article

(This article belongs to the Special Issue Explainable Artificial Intelligence (XAI) in Medical Imaging)

► Show Figures

Figure 1

17 pages, 7610 KiB

Open AccessArticle

Comprehensive Analysis of Chronic Low Back Pain: Morphological and Functional Impairments, Physical Activity Patterns, and Epidemiology in a German Population-Based Cross-Sectional Study

by Bernhard Ulrich Hoehl, Nima Taheri, Lukas Schönnagel, Luis Alexander Becker, Lukas Mödl, Sandra Reitmaier, Matthias Pumberger and Hendrik Schmidt

Bioengineering 2025, 12(8), 878; https://doi.org/10.3390/bioengineering12080878 - 14 Aug 2025

Viewed by 349

Abstract

Low back pain (LBP) is the leading cause of disability worldwide. While studies often focus on the relationship between magnetic resonance imaging (MRI) findings and symptoms or the link between pain and disability, comprehensive assessments that incorporate both structural and functional impairments are [...] Read more.

Low back pain (LBP) is the leading cause of disability worldwide. While studies often focus on the relationship between magnetic resonance imaging (MRI) findings and symptoms or the link between pain and disability, comprehensive assessments that incorporate both structural and functional impairments are lacking. This study prospectively includes standardized questionnaires, medical histories, clinical exams, and lumbar–pelvic MRI. Participants were grouped by pain status, physical activity, structural impairments (e.g., Pfirrmann, Krämer, Fujiwara, Meyerding), and posture/mobility deviations. Data were analyzed using the Kruskal–Wallis test. Of the 1262 participants, 392 (31%) reported chronic low back pain (cLBP), 226 (18%) had intermittent low back pain (iLBP), and 335 (27%) were pain-free. Significant differences were observed in high physical activity levels based on WHO criteria (cLBP: 79%, iLBP: 78%, no-BP(2): 86%, p = 0.020, η² = 0.008). Morphological impairments were more prevalent in cLBP (75%) and iLBP (76%) compared to no-BP(2) (55%) (p = 0.000, η² = 0.043). Functional impairments showed similar patterns (cLBP: 42%, iLBP: 51%, no-BP(2): 38%, p = 0.014, η² = 0.010). Participants with functional impairments tended to be younger. Consequently, the current classification system for diagnostics needs to incorporate alternative categories to more accurately differentiate types of back pain, which could enhance therapeutic outcomes. Full article

(This article belongs to the Special Issue Spine Biomechanics)

► Show Figures

Figure 1

20 pages, 3464 KiB

Open AccessSystematic Review

Evaluation of Surgical Protocols for Speech Improvement in Children with Cleft Palate: A Systematic Review and Case Series

by Angelo Michele Inchingolo, Gianna Dipalma, Paola Bassi, Rosalba Lagioia, Mirka Cavino, Valeria Colonna, Elisabetta de Ruvo, Francesco Inchingolo, Giuseppe Giudice, Andrea Palermo and Alessio Danilo Inchingolo

Bioengineering 2025, 12(8), 877; https://doi.org/10.3390/bioengineering12080877 - 14 Aug 2025

Viewed by 418

Abstract

Background: This systematic review investigates how different surgical techniques influence speech outcomes in children with cleft palate, focusing on the effectiveness of key palatoplasty methods and the timing of surgery on vocal function. Methods: A thorough search of the PubMed, Scopus, and Web [...] Read more.

Background: This systematic review investigates how different surgical techniques influence speech outcomes in children with cleft palate, focusing on the effectiveness of key palatoplasty methods and the timing of surgery on vocal function. Methods: A thorough search of the PubMed, Scopus, and Web of Science databases was conducted for studies published between 2014 and 2024, including clinical research reporting speech results after palatal repair, with bias assessed using the ROBINS tool. Additionally, two clinical cases are presented to demonstrate the practical application of the surgical approaches. Results: Analysis of fourteen studies revealed that modified Z-plasty and V-Y procedures enhance soft palate mobility and reduce hypernasality, although they require advanced surgical skills. Early closure of the hard palate, performed within the first year of life, was linked to improved consonant articulation compared to later surgeries. No significant differences were found between single-stage and two-stage repairs, but surgeon experience emerged as a crucial factor influencing outcomes. Conclusions: Overall, both the surgical technique selected and the timing of intervention play important roles in optimizing speech development in children affected by cleft palate. Full article

(This article belongs to the Special Issue New Tools for Multidisciplinary Treatment in Dentistry, 2nd Edition)

► Show Figures

Figure 1

15 pages, 1607 KiB

Open AccessArticle

Efficacy of Cross-Linked Collagen Membranes for Bone Regeneration: In Vitro and Clinical Studies

by Se-Hoon Baek, Byoung-Eun Yang, Sang-Yoon Park, Sung-Woon On, Kang-Min Ahn and Soo-Hwan Byun

Bioengineering 2025, 12(8), 876; https://doi.org/10.3390/bioengineering12080876 - 14 Aug 2025

Viewed by 434

Abstract

This study aimed to evaluate the efficacy of cross-linked collagen membranes. Two types of collagen membranes were compared: a non-cross-linked collagen membrane (group A) and a cross-linked (group B) collagen membrane. In the in vitro study, the degradation rate in the presence of [...] Read more.

This study aimed to evaluate the efficacy of cross-linked collagen membranes. Two types of collagen membranes were compared: a non-cross-linked collagen membrane (group A) and a cross-linked (group B) collagen membrane. In the in vitro study, the degradation rate in the presence of collagenase, the tear strength of the membranes, and the cytotoxicity of the cross-linked collagen membrane were evaluated. A total of 57 participants with cystic defects were randomized to undergo guided bone regeneration (GBR) using either membrane. Graft volume and new bone formation were measured by cone-beam computed tomography after 6 months of follow-up. In vitro findings revealed that the cross-linked collagen membrane retained more than 20% of its relative weight after 12 h. Meanwhile, the non-cross-linked collagen membrane exhibited complete degradation after 6 h. Clinically, no significant differences were observed between the groups in terms of graft resorption, new bone formation, and overall bone regeneration. These results indicate that cross-linking has comparable biocompatibility and enhances physical properties, including tear strength and resistance to degradation. However, clinical outcomes related to bone regeneration were not significantly different between cross-linked and non-cross-linked collagen membranes. Further research is warranted to determine the benefits of cross-linked collagen membranes in GBR procedures. Full article

(This article belongs to the Special Issue Innovative Biomaterials and Advanced Technologies in Bone Tissue Regeneration)

► Show Figures

Figure 1

12 pages, 2110 KiB

Open AccessArticle

Effect of Porcine-Derived Collagen Membrane Crosslinking on Intraoral Soft Tissue Augmentation: A Canine Model

by Blaire V. Slavin, Vasudev Vivekanand Nayak, Zachary M. Stauber, Quinn T. Ehlen, Joseph P. Costello II, Orel Tabibi, Justin E. Herbert, Ricky Almada, Sylvia Daunert, Lukasz Witek and Paulo G. Coelho

Bioengineering 2025, 12(8), 875; https://doi.org/10.3390/bioengineering12080875 - 14 Aug 2025

Viewed by 408

Abstract

Peri-implant disease and gingival recession may be partially attributed to inadequate keratinized tissue. Soft tissue augmentation procedures utilizing non-autologous biomaterials, such as porcine-derived collagen membranes, have been gaining prominence and exogenous crosslinking is being actively investigated to improve the collagen membrane’s stability and [...] Read more.

Peri-implant disease and gingival recession may be partially attributed to inadequate keratinized tissue. Soft tissue augmentation procedures utilizing non-autologous biomaterials, such as porcine-derived collagen membranes, have been gaining prominence and exogenous crosslinking is being actively investigated to improve the collagen membrane’s stability and potential for keratinized tissue gain. The aim of this preclinical study was to evaluate the performance of a novel, crosslinked porcine collagen membrane (Zderm^TM, Osteogenics Biomedical, Lubbock, TX, USA) relative to an established, commercially available, non-crosslinked counterpart (Mucograft^®, Geistlich Pharma North America Inc., Princeton, NJ, USA) in a canine mandibular model. Bilateral split-thickness mucosal defects were created in adult beagles (n = 17), with each site receiving one membrane. Qualitative and quantitative histomorphometric analyses of groups were performed after 4, 8, and 12 weeks of healing and compared to unoperated, positive controls from the same subject. No significant differences in membrane presence were observed between Zderm^TM and Mucograft^® at 4, 8, and 12 weeks of permitted healing (p > 0.05). Similarly, the average keratinized tissue (KT) length between Zderm^TM and Mucograft^® groups was statistically equivalent across all healing times (p > 0.05). However, qualitative histological evaluation revealed greater rete ridge morphology amongst defects treated with Zderm^TM in comparison to Mucograft^®. Nevertheless, both membranes exhibited excellent biocompatibility and are well-suited for soft tissue augmentation procedures in the oral cavity. Full article

(This article belongs to the Special Issue Recent Progress in Craniofacial Regeneration)

► Show Figures

Figure 1

14 pages, 2890 KiB

Open AccessArticle

Automatic 3D Tracking of Liver Metastases: Follow-Up Assessment of Cancer Patients in Contrast-Enhanced MRI

by Sophia Schulze-Weddige, Uli Fehrenbach, Johannes Kolck, Richard Ruppel, Georg Lukas Baumgärtner, Maximilian Lindholz, Isabel Theresa Schobert, Anna-Maria Haack, Henning Jann, Martina Mogl, Dominik Geisel, Bertram Wiedenmann and Tobias Penzkofer

Bioengineering 2025, 12(8), 874; https://doi.org/10.3390/bioengineering12080874 - 12 Aug 2025

Viewed by 442

Abstract

Background: Tracking differential growth of secondary liver metastases is important for early detection of progression but remains challenging due to variable tumor growth rates. We aimed to automate accurate, consistent, and efficient longitudinal monitoring. Methods: We developed an automatic 3D segmentation and tracking [...] Read more.

Background: Tracking differential growth of secondary liver metastases is important for early detection of progression but remains challenging due to variable tumor growth rates. We aimed to automate accurate, consistent, and efficient longitudinal monitoring. Methods: We developed an automatic 3D segmentation and tracking algorithm to quantify differential growth, tested on contrast-enhanced MRI follow-ups of patients with neuroendocrine liver metastases (NELMs). The output was integrated into a decision support tool to distinguish between progressive disease, stable disease, and partial/complete response. A user study involving an expert group of seven expert radiologists evaluated its impact. Group comparisons used the Friedman test with post hoc analyses. Results: Our algorithm detected 991 metastases in 30 patients: 13% new, 30% progressive, 18% stable, and 18% regressive; the remainder were either too small to measure (15%) or merged with another metastasis in the follow-up assessment (6%). Diagnostic accuracy improved with additional information on hepatic tumor load and differential growth, albeit not significantly (p = 0.72). The diagnosis time increased (p < 0.001). All radiologists found the method useful and expressed a desire to integrate it in existing diagnostic tools. Conclusions: We automated segmentation and quantification of individual NELMs, enabling comprehensive longitudinal analysis of differential tumor growth with the potential to enhance clinical decision-making. Full article

(This article belongs to the Special Issue AI-Driven Imaging and Analysis for Biomedical Applications)

► Show Figures

Figure 1

16 pages, 1932 KiB

Open AccessArticle

2.5D Deep Learning and Machine Learning for Discriminative DLBCL and IDC with Radiomics on PET/CT

by Fei Liu, Wen Chen, Jianping Zhang, Jianling Zou, Bingxin Gu, Hongxing Yang, Silong Hu, Xiaosheng Liu and Shaoli Song

Bioengineering 2025, 12(8), 873; https://doi.org/10.3390/bioengineering12080873 - 12 Aug 2025

Viewed by 541

Abstract

We aimed to establish non-invasive diagnostic models comparable to pathology testing and explore reliable digital imaging biomarkers to classify diffuse large B-cell lymphoma (DLBCL) and invasive ductal carcinoma (IDC). Our study enrolled 386 breast nodules from 279 patients with DLBCL and IDC, which [...] Read more.

We aimed to establish non-invasive diagnostic models comparable to pathology testing and explore reliable digital imaging biomarkers to classify diffuse large B-cell lymphoma (DLBCL) and invasive ductal carcinoma (IDC). Our study enrolled 386 breast nodules from 279 patients with DLBCL and IDC, which were pathologically confirmed and underwent ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) examination. Patients from two centers were separated into internal and external cohorts. Notably, we introduced 2.5D deep learning and machine learning to extract features, develop models, and discover biomarkers. Performances were assessed using the area under curve (AUC) and confusion matrix. Additionally, the Shapley additive explanation (SHAP) and local interpretable model-agnostic explanations (LIME) techniques were employed to interpret the model. On the internal cohort, the optimal model PT_TDC_SVM achieved an accuracy of 0.980 (95% confidence interval (CI): 0.957–0.991) and an AUC of 0.992 (95% CI: 0.946–0.998), surpassing the other models. On the external cohort, the accuracy was 0.975 (95% CI: 0.913–0.993) and the AUC was 0.996 (95% CI: 0.972–0.999). The optimal imaging biomarker PET_LBP-2D_gldm_DependenceEntropy demonstrated an average accuracy of 0.923/0.937 on internal/external testing. Our study presented an innovative automated model for DLBCL and IDC, identifying reliable digital imaging biomarkers with significant potential. Full article

(This article belongs to the Section Biosignal Processing)

► Show Figures

Graphical abstract

21 pages, 5025 KiB

Open AccessArticle

Cascaded Self-Supervision to Advance Cardiac MRI Segmentation in Low-Data Regimes

by Martin Urschler, Elisabeth Rechberger, Franz Thaler and Darko Štern

Bioengineering 2025, 12(8), 872; https://doi.org/10.3390/bioengineering12080872 - 12 Aug 2025

Viewed by 504

Abstract

Deep learning has shown remarkable success in medical image analysis over the last decade; however, many contributions focused on supervised methods which learn exclusively from labeled training samples. Acquiring expert-level annotations in large quantities is time-consuming and costly, even more so in medical [...] Read more.

Deep learning has shown remarkable success in medical image analysis over the last decade; however, many contributions focused on supervised methods which learn exclusively from labeled training samples. Acquiring expert-level annotations in large quantities is time-consuming and costly, even more so in medical image segmentation, where annotations are required on a pixel level and often in 3D. As a result, available labeled training data and consequently performance is often limited. Frequently, however, additional unlabeled data are available and can be readily integrated into model training, paving the way for semi- or self-supervised learning (SSL). In this work, we investigate popular SSL strategies in more detail, namely Transformation Consistency, Student–Teacher and Pseudo-Labeling, as well as exhaustive combinations thereof. We comprehensively evaluate these methods on two 2D and 3D cardiac Magnetic Resonance datasets (ACDC, MMWHS) for which several different multi-compartment segmentation labels are available. To assess performance in limited dataset scenarios, different setups with a decreasing amount of patients in the labeled dataset are investigated. We identify cascaded Self-Supervision as the best methodology, where we propose to employ Pseudo-Labeling and a self-supervised cascaded Student–Teacher model simultaneously. Our evaluation shows that in all scenarios, all investigated SSL methods outperform the respective low-data supervised baseline as well as state-of-the-art self-supervised approaches. This is most prominent in the very-low-labeled data regime, where for our proposed method we demonstrate

10.17 %

and

6.72 %

improvement in Dice Similarity Coefficient (DSC) for ACDC and MMWHS, respectively, compared with the low-data supervised approach, as well as

2.47 %

and

7.64 %

DSC improvement, respectively, when compared with related work. Moreover, in most experiments, our proposed method is able to greatly decrease the performance gap when compared to the fully supervised scenario, where all available labeled samples are used. We conclude that it is always beneficial to incorporate unlabeled data in cardiac MRI segmentation whenever it is present. Full article

(This article belongs to the Special Issue Artificial Intelligence-Based Medical Imaging Processing)

► Show Figures

Figure 1

19 pages, 2017 KiB

Open AccessArticle

Segmentation of Brain Tumors Using a Multi-Modal Segment Anything Model (MSAM) with Missing Modality Adaptation

by Jiezhen Xing and Jicong Zhang

Bioengineering 2025, 12(8), 871; https://doi.org/10.3390/bioengineering12080871 - 12 Aug 2025

Viewed by 689

Abstract

This paper presents a novel multi-modal segment anything model (MSAM) for glioma tumor segmentation using structural MRI images and diffusion tensor imaging data. We designed an effective multimodal feature fusion block to effectively integrate features from different modalities of data, thereby improving the [...] Read more.

This paper presents a novel multi-modal segment anything model (MSAM) for glioma tumor segmentation using structural MRI images and diffusion tensor imaging data. We designed an effective multimodal feature fusion block to effectively integrate features from different modalities of data, thereby improving the accuracy of brain tumor segmentation. We have designed an effective missing modality training method to address the issue of missing modalities in actual clinical scenarios. To evaluate the effectiveness of MSAM, a series of experiments were conducted comparing its performance with U-Net across various modality combinations. The results demonstrate that MSAM consistently outperforms U-Net in terms of both Dice Similarity Coefficient and 95% Hausdorff Distance, particularly when structural modality data are used alone. Through feature visualization and the use of missing modality training, we show that MSAM can effectively adapt to missing data, providing robust segmentation even when key modalities are absent. Additionally, segmentation accuracy is influenced by tumor region size, with smaller regions presenting more challenges. These findings underscore the potential of MSAM in clinical applications where incomplete data or varying tumor sizes are prevalent. Full article

(This article belongs to the Special Issue Artificial Intelligence-Based Medical Imaging Processing)

► Show Figures

Figure 1

20 pages, 2092 KiB

Open AccessReview

Quantitative Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) in Hepatocellular Carcinoma: A Review of Emerging Applications for Locoregional Therapy

by Xinyi M. Li, Tu Nguyen, Hiro D. Sparks, Kyunghyun Sung and Jason Chiang

Bioengineering 2025, 12(8), 870; https://doi.org/10.3390/bioengineering12080870 - 12 Aug 2025

Viewed by 691

Abstract

Quantitative dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is emerging as a valuable tool for assessing tumor and parenchymal perfusion in the liver, playing a developing role in locoregional therapies (LRTs) for hepatocellular carcinoma (HCC). This review explores the conceptual underpinnings and early investigational [...] Read more.

Quantitative dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is emerging as a valuable tool for assessing tumor and parenchymal perfusion in the liver, playing a developing role in locoregional therapies (LRTs) for hepatocellular carcinoma (HCC). This review explores the conceptual underpinnings and early investigational stages of DCE-MRI for LRTs, including thermal ablation, transarterial chemoembolization (TACE), and transarterial radioembolization (TARE). Preclinical and early-phase studies suggest that DCE-MRI may offer valuable insights into HCC tumor microvasculature, treatment response, and therapy planning. In thermal ablation therapies, DCE-MRI provides a quantitative measurement of tumor microvasculature and perfusion, which can guide more effective energy delivery and estimation of ablation margins. For TACE, DCE-MRI parameters are proving their potential to describe treatment efficacy and predict recurrence, especially when combined with adjuvant therapies. In ⁹⁰Y TARE, DCE-MRI shows promise for refining dosimetry planning by mapping tumor blood flow to improve microsphere distribution. However, despite these promising applications, there remains a profound gap between early investigational studies and clinical translation. Current quantitative DCE-MRI research is largely confined to phantom models and initial feasibility assessments, with robust retrospective data notably lacking and prospective clinical trials yet to be initiated. With continued development, DCE-MRI has the potential to personalize LRT treatment approaches and serve as an important tool to enhance patient outcomes for HCC. Full article

(This article belongs to the Special Issue Recent Trends and Future Directions in Biosignal Processing and Applications: State-of-the-Art Reviews)

► Show Figures

Graphical abstract

17 pages, 4171 KiB

Open AccessArticle

Effects of Aging on Motor Unit Properties in Isometric Elbow Flexion

by Fang Qiu, Xiaodong Liu and Chen Chen

Bioengineering 2025, 12(8), 869; https://doi.org/10.3390/bioengineering12080869 - 12 Aug 2025

Viewed by 439

Abstract

This study investigates age-related differences in motor unit (MU) properties and neuromuscular control during isometric elbow flexion across the human lifespan. High-density surface electromyography (sEMG) was recorded from the biceps brachii of 44 participants, divided into three groups: Child (8–14 years), Adult (20–40 [...] Read more.

This study investigates age-related differences in motor unit (MU) properties and neuromuscular control during isometric elbow flexion across the human lifespan. High-density surface electromyography (sEMG) was recorded from the biceps brachii of 44 participants, divided into three groups: Child (8–14 years), Adult (20–40 years), and Elder (65–80 years). MU spike trains were extracted noninvasively by sEMG decomposition. Then the discharge rate, MU action potential (MUAP) morphology, recruitment threshold, and common neural drive were quantified and compared across age groups. This study provides novel insights into force tracking performance, revealing that both children and elders exhibit higher errors compared to young adults, likely due to immature or declining motor control systems. Significant differences in MU discharge patterns were observed across force levels and age groups. Children and elders displayed lower MU discharge rates at low force levels, which increased at higher forces. In contrast, adults demonstrated higher MU action potential peak-to-peak amplitudes (PPV) and recruitment thresholds (RTs), along with steeper PPV-RT slopes, suggesting a narrower RT range in children and older adults. Principal component analysis revealed a strong correlation between common neural drive and force across all groups, with neural drive being weaker in elders. Overall, young adults exhibited the most efficient and synchronized MU control, while children and older adults showed distinct deviations in discharge intensity, recruitment strategies, and neural synergy. These findings comprehensively characterize MU adaptations across the lifespan, offering implications for developmental neurophysiology and age-specific neuromuscular diagnostics and interventions. Full article

(This article belongs to the Special Issue Musculoskeletal Function in Health and Disease)

► Show Figures

Figure 1

21 pages, 2629 KiB

Open AccessArticle

From Pixels to Precision—A Dual-Stream Deep Network for Pathological Nuclei Segmentation

by Rashid Nasimov, Kudratjon Zohirov, Adilbek Dauletov, Akmalbek Abdusalomov and Young Im Cho

Bioengineering 2025, 12(8), 868; https://doi.org/10.3390/bioengineering12080868 - 12 Aug 2025

Viewed by 472

Abstract

Segmenting cell nuclei in histopathological images is an extremely important process for computational pathology, affecting not only the accuracy of a disease diagnosis but also the analysis of biomarkers and the assessment of cells performed on a large scale. Although many deep learning [...] Read more.

Segmenting cell nuclei in histopathological images is an extremely important process for computational pathology, affecting not only the accuracy of a disease diagnosis but also the analysis of biomarkers and the assessment of cells performed on a large scale. Although many deep learning models can take out global and local features, it is still difficult to find a good balance between semantic context and fine boundary precision, especially when nuclei are overlapping or have changed shapes. In this paper, we put forward a novel deep learning model named Dual-Stream HyperFusionNet (DS-HFN), which is capable of explicitly representing the global contextual and boundary-sensitive features for the robust nuclei segmentation task by first decoupling and then fusing them. The dual-stream encoder in DS-HFN can simultaneously acquire the semantic and edge-focused features, which can be later combined with the help of the attention-driven HyperFeature Embedding Module (HFEM). Additionally, the dual-decoder concept, together with the Gradient-Aligned Loss Function, facilitates structural precision by making the segmentation gradients that are predicted consistent with the ground-truth contours. On various benchmark datasets like TNBC and MoNuSeg, DS-HFN not only achieves better results than other 30 state-of-the-art models in all evaluation metrics but also is less computationally expensive. These findings indicate that DS-HFN provides a capability for accurate nuclei segmentation, which is essential for clinical diagnosis and biomarker analysis, across a wide range of tissues in digital pathology. Full article

(This article belongs to the Special Issue Medical Imaging Analysis: Current and Future Trends)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Bioengineering, Volume 12, Issue 8 (August 2025) – 111 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI