Bioengineering

Background/PurposeDiagnosing wrist ligament injuries is challenging; early detection and treatment are important to prevent osteoarthritis progression. Interosseous proximity maps, a proxy measure for joint space, can be generated from volumetric imaging data and may provide important information about wrist health. Artificial intelligence (AI) could enhance accuracy of noninvasive diagnosis based on imaging-derived metrics. This work demonstrates feasibility of AI training using synthetic proximity map data generated from finite element models (FEMs). MethodsPersonalized wrist FEMs for two asymptomatic participants were created from four-dimensional computed tomography-derived anatomic and kinematic data. Monte Carlo sampling varied 22 ligament material properties and simulated 7,500 unique injury scenarios generating 9,000,000 labeled red, green, and blue (RGB) images of interosseous proximity vector fields from FEM-derived motions. Images were associated with 17 descriptive metrics, including gross wrist angles and bone surface pairs, and used to develop mixed-input convolutional neural networks (CNNs). Model performance was evaluated for identifying specific ligament injuries. ResultsAverage area under receiver operating characteristic curve (AUROC) for CNNs was 0.757 across all injury types and kinematics. In a subset with clinically-relevant functional angles, the average AUROC was 0.824. Best-performing individual ligament AUROCs ranged from 0.807 to 0.999. Sensitivities and specificities exceeded 0.99 for some ligament injury simulations under specific wrist angles and bone surface pairs. ConclusionThis study demonstrates the feasibility of using synthetic data from FEMs to train AI models for classifying wrist ligament injuries. Proximity-based RGB images may be a relevant biomarker of ligamentous injury.

Background: Midurethral sling placement is often performed during prolapse repair to treat or prevent stress urinary incontinence. However, some women experience persistent or new-onset stress or urgency urinary incontinence after surgery. It is unclear how prolapse repair, with or without a concomitant midurethral sling, alters urethral morphology and support, and how these changes relate to urinary continence outcomes. Objectives: To compare postoperative urethral morphology (dimensions, angles, shape) and support (position, mobility) after transvaginal prolapse repair with vs without a concurrent midurethral sling, and to explore associations between postoperative urethral characteristics and urinary outcomes (stress, urgency symptoms). Study Design: This ancillary analysis used magnetic resonance imaging and urinary outcome data from the Defining Mechanisms of Anterior Vaginal Wall Descent Study conducted across 8 clinical sites within the United States Pelvic Floor Disorders Network. Eighty-two women (median age, 65 years) underwent transvaginal prolapse repair (vaginal mesh hysteropexy or vaginal hysterectomy with uterosacral ligament suspension) with or without a concurrent midurethral sling between April 2013 and February 2015. Postoperative imaging at rest and during strain was performed 30-42 months after surgery (or earlier if they chose reoperation) between June 2014 and May 2018. Prolapse recurrence, defined as descent beyond the vaginal introitus during strain, was recorded. The urethra was segmented from postoperative scans to create 3-dimensional models for measuring urethral diameters, length, surface area, volume, angles, shape (principal component scores from a statistical shape model), position, and mobility (rest-to-strain displacement). Preoperative and 24-48-month postoperative urinary continence outcomes were assessed using validated questionnaires: the Urogenital Distress Inventory, Urinary Impact Questionnaire, and the Incontinence Severity Index. Comparisons of urethral and urinary outcomes by (1) midurethral sling and (2) stress urinary incontinence were made using Wilcoxon rank-sum tests, principal component analysis, and multivariate models as appropriate. Associations between urethral and urinary outcomes were evaluated with Spearmans rank correlation. Results: Forty-six women (22 hysteropexy, 24 hysterectomy) were in the sling group, and 36 (19 hysteropexy, 17 hysterectomy) were in the no-sling group. Among the 48 women without prolapse recurrence (28 sling, 20 no-sling), those with a sling (vs without) had larger urethral dimensions (all P<.03), a more anterior-superior position of the proximal urethra (indicating better bladder neck support) (P=.04), a straighter urethral shape (P=.006), and reported less bothersome postoperative stress incontinence (P=.02). Overall, 14 women (17%) experienced postoperative stress incontinence. Stress urinary incontinence was linked to a more acute proximal urethral sagittal angle (more aligned with axial plane) (P=.01), and a lower proximal urethra position (P=.04) and mid-urethra position (P=.03). Poorer stress and urgency urinary outcomes were associated with a shorter urethral length (P=.01), a more posterior-inferior urethral position (all P<.05), increased C or S-shaped urethral concavity (P=.008; P=.006), and smaller rest-to-strain displacement of the proximal (P=.03) and distal (P=.009) urethra. Conclusions: Urethral morphology and support differed with concomitant midurethral sling (vs no sling) and stress urinary incontinence after vaginal surgery. Urethral characteristics were also associated with postoperative urinary symptoms. Urethral configuration may influence urinary outcomes and could be considered during prolapse and stress urinary incontinence repairs.

Objective: To develop and validate a robust deep-learning model capable of fine-grained phase recognition in total hysterectomy, particularly the complex periuterine dissection phase. Design: Multicentre retrospective observational study. Setting: Japan. Sample: Surgical videos (n = 764) from 43 institutions. Methods: We developed a robust and generalisable deep-learning model for surgical phase recognition in total hysterectomy, applicable to laparoscopic and robot-assisted procedures. Overall, 1,591,334 still images were annotated across nine surgical phases. A convolutional neural network (Xception architecture) was trained on 200 cases using four-fold cross-validation, with institutional separation between training and testing sets. Main outcome measures: Model performance was assessed using accuracy, precision, recall, and F1 score. Subgroup analysis and logistic regression evaluated the association between background clinical factors and recognition accuracy. Results: The model achieved an overall phase recognition accuracy of 0.78 (95% CI: 0.74--0.80), with a precision of 0.75 (95% CI: 0.72--0.78) and a recall of 0.76 (95% CI: 0.74--0.78). Performance was consistent across laparoscopic and robot-assisted procedures and across most surgical phases. Accuracy plateaued after training on 120 cases. No clinical factors significantly impacted performance. Trends toward lower accuracy were observed for cases with cervical myoma and pouch of Douglas adhesions. Conclusions: This model demonstrated high accuracy across diverse institutions and patient backgrounds. Its potential applications include surgical education, real-time intraoperative support, and training efficiency enhancement.

Acute subdural hematoma (ASDH) is a prevalent injury with high mortality and morbidity, often resulting from bridging vein (BV) disruption secondary to cortical relative motion. As a thin membrane enveloping the brain surface and anchoring BVs, the pia mater is hypothesized to play a critical mechanical role in cortical response and hence ASDH pathogenesis. Finite element (FE) head models are valuable tools to predict ASDH occurrence during impacts. However, the pia mater is often represented as an elastic material in existing FE head models, despite experimental evidence reporting its nonlinear mechanical behavior. In this study, both linear (Young's modulus of 11.5 MPa) and nonlinear (the stress-strain curve derived from pial tension tests) material models of the pia mater were implemented in one FE head model. The models were subjected to three experimental impact loadings, one of which was known to cause ASDH and two of which were not. Results demonstrated that, across all simulated impacts, the model with nonlinear pia mater properties predicted larger cortical displacements and BV responses than the linear model. For the impact with known ASDH occurrence, the predicted BV strain was 0.17 for the nonlinear model and 0.094 for the linear model, with only the former approaching the reported rupture strain range of the BV-superior sagittal sinus complex (0.29 {+/-} 0.13). These findings verified the mechanical importance of the pia mater in cortical responses and hence the prediction of ASDH, suggesting that conventional linear pia modeling might over-constrain cortical motion, leading to underestimation of BV strain and ASDH risk. The current study supported the adoption of experimentally derived nonlinear pia mater properties in FE head models to improve the reliability of ASDH prediction.

Preterm birth is the leading cause of neonatal death. Despite sustained efforts to identify high-risk women in the mid-trimester, accurate prediction remains difficult. Quantitative cervical ultrasound texture has been proposed as a predictor of spontaneous preterm birth. However, earlier models were developed in small single-centre samples and were not externally validated. We developed image-texture (Local Binary Patterns with a Random Forest), deep-learning (Vision Transformer), clinical-variable, and multimodal models to predict spontaneous preterm birth on the prospective GARBH-Ini cohort. We then externally validated our best models on an independent cohort scanned on a different ultrasound machine. Our best overall model reached an internal-test area under the receiver-operating-characteristic curve of 0.71 (95% CI 0.60, 0.82), but performed modestly at 0.52 (95% CI 0.38, 0.64) externally. The deep-learning and multimodal models did not perform better. Discrimination appeared higher in a clinically high-risk subgroup at the 34-week threshold. These estimates were imprecise because of few cases and need to be confirmed in future studies. Among the several likely reasons for the modest external performance is the heterogeneity of preterm birth. Predicting distinct preterm-birth subtypes separately, and integrating additional biomarkers and data domains, might improve model performance. Keywords: preterm birth; cervical ultrasound; prediction model; external validation; deep learning

Background Delineating the cellular origins of extracellular vesicles (EVs) enables the detection of clinically relevant changes in dynamic and complex tissues, such as the endometrium, which are not characterizable through single biomarker assays. Transcriptome deconvolution into cellular composition using deep learning methods provides a means to explore this complexity. However, such computational methods have not been previously applied to EV bulk transcriptomes, and their efficacy in profiling EV population changes and concordance to tissue throughout the menstrual cycle remains unknown. Methods This observational cross-sectional study utilized a deconvolutional generative deep learning algorithm, BulkTrajBlend, trained on a comprehensive human endometrial single-cell RNA sequencing (scRNA-seq) atlas. The model was applied to deconvolve paired bulk transcriptomes from endometrial tissue and uterine fluid EVs (UF-EVs) across the proliferative (P, n=4), early-secretory (ES, n=5), mid-secretory (MS, n=5), and late-secretory (LS, n=5) phases from healthy, fertile women. To validate generalizability, independent UF-EV datasets (ES, n=12; MS, n=12) obtained via different laboratory protocols were included. Deconvolved pseudo-single-cell (pSC) profiles from UF-EV data were subsequently integrated with Visium spatial transcriptomics slides of human endometrium (P, n=2; MS, n=4; ES, n=2). Results We developed a foundation model-based approach utilizing self-supervised learning to determine the cellular origin of EVs from their transcriptomic profiles. By mapping the generated pSC profiles to spatial transcriptomic data, we evaluated spatial origins of EVs. The statistical analysis demonstrated that UF-EV transcriptome deconvolution reflects the dynamic changes in the cellular composition of endometrial tissue across the menstrual cycle phases. The ability to distinguish accurately between proliferative and decidualizing menstrual cycle phases (ROC-AUC = 0.98) using cellular profile of deconvoluted UF-EVs transcriptome enables non-invasive profiling of endometrial tissue. Conclusions Our findings indicate the feasibility of determining endometrial tissue cellular composition using UF-EV transcriptomics. This methodology enables refined, non-invasive endometrial testing, avoiding invasive biopsy procedures. Based on deconvolution results, we are able to correlate UF-EV content to tissue, and distinguish between menstrual cycle phases. These results build toward a multifactorial screening method for abnormalities within the endometrium.

BackgroundFrom the radiographic perspective, the septum pellucidum (SP) and septum verum (SV) Complex (SPVC) has been tacitly understood. Microdissection and diffusion tensor imaging (DTI) have now well established that they are not mere membranes but contain septal nuclei and nerve fibers; the Superior (SF) and Inferior fascicles (IF) forming the SP, and precommissural fibers of the fornix (PrCFx) in SV. ObjectiveWe aimed to delineate the topography of normal and abnormal SPVC using ultrasound (US), T2-weighted magnetic resonance imaging (MRI), and DTI in fetuses and provide an algorithm for prenatal diagnosis and evaluation of septopreoptic holoprosencephaly (SPrH). MethodsTwenty-nine fetuses included in the study were divided based on US into Group 1 (five of 29): normal Cavum Septum Pellucidum (CSP) on axial transthalamic (aTTP) and transventricular (aTVP) planes; Group 2 (eleven of 29): non-visualization of the SP in aTVP, coronal transcaudate plane (cTCP) and beyond; Group 3 (three of 29): single septum in aTVP; Group 4 (ten of 29): small /echogenic CSP in aTTP and aTVP. ResultsAll three fascicles forming the SPVC were demonstrated in all cases prenatally and/or postnatally on US, MRI and DTI. All fetuses in Groups 2 to 4 showed an abnormal hypointense band bridging the region of septal and/or preoptic nuclei on T2-weighted fetal and postnatal MR, suggestive of SPrH. ConclusionThis study contributes to understanding the topography of normal and abnormal SPVC by prenatal US, MRI, and DTI. Based on this understanding, we outline an algorithm for prenatal diagnosis and evaluation of SPrH. HighlightsO_LIFetal Septum pellucidum/verum complex (SPVC) contain septal nuclei and 3 nerve fiber groups: Superior fascicle, Inferior fascicle and Precommissural fornix C_LIO_LIThese are seen on ultrasound, T2-weighted MRI and Diffusion tensor imaging in cases with both normal and abnormal cavum septum pellucidum (CSP). C_LIO_LIHypointense band in the septopreoptic region on T2-weighted MRI in fetuses with abnormal CSP are potential markers of septopreoptic holoprosencephaly C_LIO_LIRecognition of this entity may help in prenatal counselling and prognosis C_LI

Introduction: Critical-sized bone defects cannot heal spontaneously, requiring additional, often burdensome, treatment. Thus, various synthetic substitute materials have been investigated regarding their treatment capacity. Poly-L-lactic acid (PLLA) and polyglycolic acid (PGA) have emerged as promising biodegradable scaffold materials. The addition of inorganic materials such as calcium carbonate (CC) has also been shown to be advantageous. This study investigates the effect on bone regeneration of PLLA-PGA-CC scaffolds in critical-sized bone defects over a two-year observation period using sheep as an animal model. Methods: Critical-sized mandible angle defects were created in twelve female merino sheep. Mandibular defects were reconstructed with PLLA-PGA-CC scaffolds in four sheep, while the remaining eight served as negative control (defects left empty). The scaffolds were manufactured using computer-aided design and manufacturing, incorporating an interconnected porous structure and fixated with polyether ether ketone cages. Bone regeneration was evaluated using computed tomography (CT) imaging at 3, 12, and 24 months postoperatively. Bone volume was assessed quantitatively. Additionally, a histological analysis was performed. Results: Surgical procedures were successful and without major complications. CT assessment showed more bone regeneration in the scaffold group (mean volume: 7,472 mm3) than in the control group (4,168 mm3, p = 0.1) at 24 months postoperatively. Resorption of the scaffolds and formation of compact lamellar bone tissue were confirmed by histological analysis. However, the osteoconductive properties of the scaffolds were limited, with only minimal ingrowth of bone tissue into the porous structure. In both groups, fibrous tissue infiltration and the formation of cyst-like cavities in the defect region were observed. Conclusion: PLLA-PGA-CC scaffolds were found to be biocompatible and enhanced bone regeneration compared to the control group. Due to fibrous tissue infiltration and the lack of osteoconductivity, the suitability of the material for critical-sized bone defect reconstruction is limited.

Accurate and comprehensive evaluation of three-dimensional (3D) pelvic-floor-muscle (PFM) strength distributions are highly expected to play a crucial role in clinical early diagnosis and physiotherapy precise treatment of womens pelvic-floor-dysfunction (PFD). However, clinically existing PFM-evaluating methods merely assess a rough and synthetical PFM strength from the whole vagina muscle tunnel, seriously restricting the development of PFD-related diagnostic methodologies and therapeutic interventions. Here, 3D complex female PFM-strength distributions have been accurately detected by developing a portable multi-channel PFM-pressure dynamic measuring system. Clinical trials demonstrate the superiority in high PFM-strength sensitivity and 3D spatial resolution, offering the opportunities to imply specific deficits in personalized PFM functions and customized interventions accordingly. Significantly, various subclinical PFM abnormalities can be identified by the 3D accurate PFM-strength distributions, which is not possible using traditional PFM-evaluating methods, providing a visual biomechanical foundation for clinical early diagnosis and physiotherapy precise treatment of PFD. Combined with the additional advance in physical comfortability, patients-friendliness universality, and stability without motion artifact achieved by the novel designed vaginal probe, this proof of concept research holds the promise for paradigm revolution in PFM pathological research, and promotes the transformation of clinical pelvic-floor medicine from empirical medicine to data-driven precision medicine.

Patellar osteochondral allograft (OCA) transplantation is widely used to treat large full-thickness cartilage defects, yet long-term failure and reoperation rates remain high. Although surface congruity and osseous integration are emphasized clinically, cartilage thickness and mechanical compatibility between donor and recipient are not considered. Our previous work suggests that cartilage thickness mismatch can amplify local deformation at the graft boundary, potentially compromising graft longevity. This study investigates how combined mismatches in cartilage thickness and mechanical properties influence the local strain environment at the patellar OCA interface. Simplified two-dimensional axisymmetric finite element models of patellar OCA repair were developed in ABAQUS. Donor-to-recipient cartilage thickness ratios ranging from 0.33 to 3.25 were evaluated together with donor-recipient Youngs modulus mismatches (2.5-7.0 MPa). Cartilage was modeled using homogeneous linear elastic and functionally graded material formulations to account for depth-dependent stiffness. A compressive pressure of 1.0 MPa was applied to represent patellofemoral joint loading, and peak compressive and shear strains were quantified at the graft boundary. Cartilage thickness mismatch produced localized high-strain regions (HSR) of compressive and shear strain at the donor-recipient interface that were absent in thickness-matched constructs. Strain amplification increased with both thickness and mechanical property mismatch. Compressive strain exhibited directional asymmetry, with donor-side-thicker configurations producing greater amplification than recipient-side-thicker configurations. Incorporating depth-dependent cartilage stiffness reduced peak strain magnitudes but did not eliminate mismatch-driven strain amplification. These findings demonstrate that cartilage thickness and mechanical disparity can create HSR at the patellar OCA graft boundary that may predispose grafts to impaired integration and long-term failure.

Tissue-level mechanical stimuli are primary drivers of tissue adaptation and can be optimised during conservative treatments to improve treatment outcomes for many highly prevalent musculoskeletal conditions. Current laboratory-based technologies limit our ability to connect conservative interventions such as exercise and movement modification with muscle, joint, and tissue-level mechanics, in natural environments. We introduce a physics-informed neural network (PINN) to estimate clinically relevant biomechanics from smart garments. By accounting for physiological dynamics of neural activation and muscle contraction, the PINN accurately predicted hip joint angles (RMSE <6 degrees), moments (RMSE 0.12 N*m/kg to 0.30 N*m/kg), and joint forces (RMSE 6 to 16%) from three inertial measurement units and four electromyographic sensors. We demonstrated that the trained PINN can be combined with a smart garment to estimate hip biomechanics, in real-time, during a gait retraining intervention aimed at modifying joint loading to treat hip osteoarthritis. The developed PINN and smart garment system may be adapted and generalised for personalised management or rehabilitation of a broad range of musculoskeletal diseases and injuries, in clinical, home, workplace, and sporting environments.

Total knee replacement restores mobility in patients with advanced osteoarthritis, yet many individuals still experience limited ability to perform high-flexion tasks such as squatting. Current preoperative planning relies on static imaging and cannot predict how different implant alignment choices will affect postoperative dynamic function. This study developed a predictive simulation framework that uses bi-level inverse optimal control to link preoperative implant alignment directly to expected postoperative squat kinematics. Subject-specific musculoskeletal models were constructed for six total knee replacement patients using experimental squat data. Bi-level inverse optimal control was applied to identify both individualised and group-level cost functions. The individualised setting provided subject-specific accuracy, while the group-level setting derived a single group-level cost function as an initial step toward preoperative use without requiring postoperative motion data. The individualised setting reproduced experimental trajectories with low errors across all joints (mean apex difference 1.53{degrees}, root-mean-square error 5.15{degrees}, normalised root-mean-square error 11.15%, Pearson correlation 0.96). The group-level setting yielded higher but acceptable errors (mean apex difference 5.70{degrees}, root-mean-square error 6.75{degrees}, normalised root-mean-square error 17.53%, Pearson correlation 0.95) while preserving the general pattern and phasing of the motion. Squat depth emerged naturally from the optimisation rather than being prescribed. This framework may provide a basis for future quantitative tools to evaluate how implant alignment choices influence postoperative squat performance, potentially improving functional outcomes in total knee replacement. These results suggest that the proposed IOC framework can reproduce key features of post-TKR squat kinematics, but further out-of-sample validation is required before it can be used for preoperative prediction or translated into tools aimed at improving functional outcomes in total knee replacement.

ObjectiveTo systematically review automated nerve morphometry tools and independently benchmark their performance on independent optic nerve datasets. DesignSystematic review and comparative benchmarking study. ControlsBenchmarking was performed using paraphenylenediamine-stained mouse (n = 85) and rat (n = 44) optic nerve images with manually annotated axon counts as ground truth. MethodsPublished studies describing automated or semi-automated neural tissue morphometry tools were identified through systematic searches of PubMed, Embase, and Scopus through January 2026 following PRISMA guidelines. Data extraction covered 70 fields across tool capabilities, imaging modality, species, automation level, and validation approach. Eighteen eligible tools (8 deep learning [DL], 10 classical computer vision [CV]) were benchmarked on both mouse and rat independent datasets. Main Outcome MeasuresPerformance was assessed by mean absolute percentage error (MAPE), Pearson correlation, and median predicted-to-ground-truth ratio. Tools were ranked per image and compared using Friedman tests with Nemenyi post-hoc analysis. ResultsSeventy-one studies met inclusion criteria, spanning from 1999 to 2026. Deep learning methods represented 38% (27/71) of studies, increasing from 0% before 2017 to over 55% of publications after 2020. Axon counting was the most common output (73%, 52/71), while only 35% (25/71) reported g-ratio. Among benchmarked tools, Marina (CV, 2010) achieved the lowest average MAPE (32.9%). The top five tools (MAPE ranging from 32.9 to 44.8%) included both CV and DL methods and were statistically indistinguishable by Friedman-Nemenyi analysis (p > 0.05). Performance varied substantially across datasets: AxonJ (CV) achieved the second best MAPE on rat images (27.7%) but the worst on mouse images (438.6%). ConclusionsNo single tool demonstrated consistently superior performance across both datasets. Classical and deep learning approaches achieved comparable accuracy for axon counting. Tool selection should be guided by target species, tissue preparation protocol, and desired morphometric outputs. This systematic review and independent benchmarking study provide an evidence base for tool selection in optic nerve research.

Xenotransplantation using genetically engineered pig organs offers a promising solution to the shortage of donor organs for life-saving transplants. However, human preformed antibodies against unknown pig xenoantigens remain a significant barrier to successful xenotransplantation. Current methods for characterizing these antibodies or xenoantigens are limited to cellular-level crossmatch assays. In this study, we developed a novel approach to identify pig xenoantigens, including peptide and glycopeptide epitopes that react with human preformed antibodies. First, human preformed antibodies against xenoantigens were enriched from plasma using immobilized pig kidney proteins. The enriched antibodies were then immobilized and used to isolate pig kidney proteins, peptides, and intact glycopeptides, followed by liquid chromatography-tandem mass spectrometry analysis. This dual-level approach identified 221 peptides corresponding to 153 proteins, with a significant enrichment of plasma membrane and extracellular proteins. Notably, 11 peptides were unique to pig sequences, suggesting their potential role in driving xenogeneic immune responses. Glycoproteomic analysis identified 122 intact glycopeptides, predominantly complex/hybrid glycoforms and Neu5Gc-containing glycans. Our method effectively identifies peptides and intact glycopeptides reactive to human preformed antibodies, providing critical insights for discovering xenoantigens. These findings could guide genetic engineering strategies and enhance recipient candidate screening for xenotransplantation, ultimately increasing the feasibility and success of xenogeneic organ transplantation.

PurposeTo characterize the frequency-dependent bioimpedance properties of major ocular tissues in intact ex vivo porcine eyes under simulated surgical conditions and evaluate tissue separability at discrete frequencies. MethodsBioimpedance spectra were acquired from sclera, corneal epithelium, iris, lens, vitreous, and retina in intact ex vivo porcine eyes using a two-electrode probe and a precision LCR meter over 5 kHz to 1 MHz. Measurements were obtained under balanced salt solution and ophthalmic viscosurgical device conditions. Probe-tissue contact was verified by microscope visualization and optical coherence tomography. Tissue separability at 5, 50, 100, and 900 kHz was evaluated using global and pairwise statistical comparisons, effect sizes, and ROC-based separability metrics. Robotic-stabilized and handheld measurements were also compared. ResultsOcular tissues demonstrated distinct, frequency-dependent impedance magnitude distributions. Across sampled frequencies, 60% to 80% of tissue pairs showed significant differences after multiplicity correction. Median pairwise effect sizes ranged from Cohens d = 0.48 at 5 kHz to 1.04 to 1.06 at 50 to 100 kHz. Median ROC-based separability was 0.91 at 5 kHz and 0.76 to 0.77 at 50 to 900 kHz. Robotic-stabilized measurements showed lower variance than handheld measurements, although tissue-specific impedance ranges and frequency-dependent trends were preserved across acquisition modes. ConclusionsMajor ocular tissues exhibit reproducible, frequency-dependent bioimpedance signatures in intact ex vivo eyes under simulated surgical preparation. These findings establish a physiologically relevant ocular impedance reference dataset and support bioimpedance as a complementary modality for tissue differentiation in ophthalmic microsurgery.

Manufacturer-defined signal-strength indices are frequently employed as quality benchmarks for automated optical coherence tomography analysis, yet their empirical relationship with deep learning segmentation accuracy remains unclear. Because these metrics were originally developed for conventional image-processing pipelines, their ability to predict modern model-based segmentation accuracy has not been empirically validated. To address this gap, we evaluated the Heidelberg Spectralis Q-score against U-Net segmentation performance across 5,047 B-scans from 103 eyes for three anatomical boundaries of the posterior segment of the eye: the Ellipsoid Zone (EZ), Bruch's Membrane (BM), and Choroid Outer Boundary (COB). Alongside standard boundary agreement metrics (MAE, MSE, Dice Similarity Coefficient), we adapted the Earth Mover's Distance (EMD) from optimal transport theory as a boundary evaluation metric. Unlike column-wise averages, EMD quantifies boundary agreement as a 2-D geometric displacement, directly measuring residual spatial displacement between the model segmented boundary and the ground-truth boundary. Our results demonstrate that the Q-score - originally designed to gate image-processing-based automated analysis - is a poor predictor of deep learning boundary segmentation accuracy, with explained variance (R2) failing to exceed 1.4% across all three boundaries. We further observed a monotonically increasing error hierarchy with anatomical depth (EZ < BM < COB), consistent across metrics, which is unexplained by the signal strength. At the COB, correlations were paradoxically positive, explained by a B-scan-level mediation chain: higher Q-scores correspond to greater choroidal thickness (r=0.113, {rho}=0.158), which in turn predicts higher COB segmentation error (r=0.165, {rho}=0.191) - a localization difficulty that global signal strength cannot capture. Collectively, these findings challenge the implicit assumption that signal-strength-based quality thresholds are a reliable proxy for deep learning model performance, and motivate a shift toward task-specific acquisition quality criteria calibrated to model performance rather than signal interpretability.

Brain tumor detection using Magnetic Resonance Imaging (MRI) remains a challenging task due to tumor heterogeneity and imaging variability. This paper presents a novel hybrid Deep Convolutional Neural Network-Whale Optimization Algorithm (DCNN-WOA) framework for automated brain tumor detection and classification. The proposed method consists of four main stages: MRI data preprocessing and augmentation, deep feature extraction using multi-layer Convolutional Neural Networks (CNN), feature selection and hyperparameter optimization via the Whale Optimization Algorithm (WOA), and final classification with comprehensive performance evaluation. By jointly optimizing deep features and training parameters, the framework effectively reduces feature redundancy, accelerates convergence, and enhances model generalization. Experimental results on a publicly available MRI dataset demonstrate that the DCNN-WOA model outperforms conventional CNN and state-of-the-art Deep Learning (DL) architectures, achieving an accuracy of 97.8%, sensitivity of 96.4%, specificity of 98.1%, and F1-score of 97.2%. The practical impact of this approach makes it a promising solution for real-time clinical decision-support systems in neuroimaging.

BackgroundBone graft biomaterials play a critical role in bone regeneration by influencing osteoblast differentiation and mineralization. However, comparative data regarding the osteogenic potential of commonly used graft materials under standardized conditions remain limited. Method and materialIn this in vitro experimental study, osteoblast-like cells (MG-63) were cultured with four bone graft materials, including Bio-Oss, Cerasorb, Bio-Tiss Cerabone, and Pro Osteon. The relative mRNA expression of osteogenic markers (COL1 and OPN) was evaluated at 1, 7, 14, and 21 days using real-time PCR. Alkaline phosphatase (ALP) activity and mineralization capacity were also assessed using colorimetric assay and Alizarin Red staining. Data were analyzed using one-way ANOVA and Tukey post hoc test (P < 0.05). ResultsSignificant differences were observed among the tested materials across all evaluated parameters. Bio-Oss and Cerasorb demonstrated higher gene expression levels and ALP activity compared to Bio-Tiss Cerabone and Pro Osteon (P < 0.05). Mineralization analysis showed significantly greater calcium deposition in the Bio-Oss and Cerasorb groups, whereas Pro Osteon consistently exhibited the lowest osteogenic performance. ConclusionBone graft biomaterials significantly influence osteogenic activity in osteoblast-like cells. Bio-Oss and Cerasorb showed superior osteogenic potential, while Pro Osteon demonstrated weaker performance. These findings highlight the importance of material properties in optimizing bone regeneration.

Purpose: To investigate how artificial intelligence (AI) systems detect referrable diabetic retinopathy (DR) from retinal photographs by analysing heatmap patterns and determining their overlap with DR features. Methods: Fifty-four AI systems were developed using 27 backbone architectures, with each implemented as both binary-referable and multi-class grading models based on the International Clinical Diabetic Retinopathy (ICDR) grading scale. Models were trained on images from DDR, BRSET and Kaggle datasets. After training, each model analysed 749 images with DR feature annotations, with Grad-CAM heatmaps generated and compared to pixel-level annotations of microaneurysms, haemorrhages, exudates, cotton wool spots, venous beading, intraretinal microvascular abnormalities and neovascularisation. Results: All models achieved acceptable predictive performance (AUROC >0.8 for most architectures). Heatmap analysis revealed consistent attention to the macular region with relative neglect of the optic disc. Exudates and cotton wool spots were highlighted most frequently by the heatmaps, with venous beading and neovascularisation at the disc showing poor overall coverage for binary referable classifiers. Models grading per the ICDR scale demonstrated high coverage for all features. Substantial variability was observed between architectures, suggesting different feature detection capabilities. Interestingly, the heatmap analysis indicated that the models were using different logic to the ICDR grading scale definitions. Conclusion: AI models do not uniformly rely on all DR features when detecting referable DR, limiting their predictive performance in unusual presentations. Heatmap aggregation analysis provides a scalable method for analysing model behaviour, allowing strengths and weaknesses to be identified. These findings may help improve clinician's trust and acceptance of AI.

Accurate evaluation of extracellular matrix (ECM) mineralization at the nano-scale is essential for establishing relevant in vitro bone models. This is particularly important with the development and increased application of three-dimensional (3D) cell models for biological research. Transmission electron microscopy (TEM) allows to perform ultra-structural analysis of cells and ECM organization, but its application in in vitro bone models remains limited, due to the potential alteration or loss of the mineral phase during sample preparation. In this study, we compared two TEM sample preparation methods - the conventional chemical fixation and the anhydrous methods - to evaluate their ability to preserve the mineralized ECM in MC3T3-E1 cells cultured as monolayers and as alginate-encapsulated bone spheroids. Chemical fixation preserved cellular ultra-structure and collagen organization, allowing for detailed assessment of cells and ECM organization. Although mineral deposits were detected and their needle-like morphology assessed, characterization of more immature deposits was partially limited by the effects of uranyl acetate and the overall sample preparation process, which could lead to alteration or loss of less stable mineral phases. The anhydrous preparation method resulted in limited preservation of cellular and ECM morphology and did not allow reliable identification of mineral deposits. When applied to spheroids, the chemical fixation method preserved the 3D architecture, collagen-rich ECM and inner mineral deposits, confirming spheroids as a relevant model for bone studies. Overall, these results highlight the need for optimized sample preparation strategies that preserve both ultra-structure and mineral components for accurate nano-scale characterization of bone mineralization.