ABME

Vision-threatening retinal damage is often detected too late, when treatment options become more limited and sight loss harder to prevent. Researchers developed Deep Decoder-Focused U-Net, an AI model designed to automatically segment four major retinal lesion types from retinal scans – soft exudates, hard exudates, microaneurysms, and hemorrhages. The model uses an asymmetric architecture with a denser decoder to better preserve fine structural details and improve identification of small, complex lesions that are difficult to segment manually. Across two major retinal imaging datasets, the approach outperformed existing state-of-the-art segmentation methods, demonstrating stronger accuracy in distinguishing multiple lesion types simultaneously. Advancing lesion detection could help clinicians identify eye disease sooner, reduce diagnostic burden, and preserve vision before irreversible damage occurs. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: DDFU-Net: A Deep Decoder-Focused U-Net Model for Retinal Lesion Segmentation 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: María Herrero-Tudela, Roberto Romero-Oraá, Gonzalo C. Gutiérrez-Tobal, @robhor, María I. López, Pedro Romero-Aroca, and @MariaGarciaGad @universidaddeva @ETSIT_UVa @GIB_UVa @universitatURV @HospitalReus 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Tracking how blood moves through the brain is critical for studying neurological disease, yet accurately measuring that flow in small-scale preclinical models has seen limited progress. To establish more reliable measures of cerebral circulation, researchers tested the accuracy of phase-contrast MRI and established reference values for major intracerebral arteries and veins in a preclinical model. The study combined controlled flow phantom testing with in vivo measurements, using semi-automatic image segmentation to quantify blood movement and pulsatility across key vessels. Results showed strong agreement between MRI-derived measurements and reference pump values, while also confirming balanced cerebral inflow and outflow and symmetrical circulation between the left and right sides of the brain. By strengthening the accuracy of preclinical brain-flow measurements, this work could help accelerate earlier detection and better treatment of neurological disease. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Characterization of Arterial and Sinus Blood Flow Dynamics Using Phase-Contrast MRI 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Sidy Fall, Kamel Abderrahim, and Olivier Baledent @UPJV_Univ 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Compression therapies can improve circulation and reduce swelling, though designing them efficiently depends on simulations that are often too slow and computationally expensive for widespread use. Taking aim at this bottleneck, researchers developed a multi-fidelity framework that simulates how external compression affects both tissue movement and blood flow. The system linked a one-dimensional blood flow model with 3D tissue mechanics and was evaluated in both simplified and subject-specific geometries. Compared with full 3D simulations, the framework: 🔹 Maintained flow-rate errors below 1% and pressure errors below 2% for most of the compression cycle 🔹 Reduced computational cost by up to 46-fold 🔹 Successfully simulated a full-cycle intermittent pneumatic compression operation over extended time scales Faster, scalable simulations could accelerate the development of more effective compression therapies and medical devices that are better tailored to physiology. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Efficient Multi-Fidelity Fluid–Structure Interaction Modeling for Pulsatile Blood Flow in Deformable Biological Tissues 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Chang Min Lee, Youngjae Choi, Kiwon Lee, Mihyun Lee, Seung-Hoon Kim, Yong-Soon Yoon, and Hyun Jin Kim @kaistpr @KAISTCoE @kaist_me 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Hernia repair failures can force people back into repeated surgeries because surgeons still lack precise tools to predict how much tension the abdominal wall can safely withstand during closure. Researchers explored a potential solution by developing a patient-specific computational model that simulated tension changes during transversus abdominis release, a procedure used in complex ventral hernia repair. Using a preoperative CT scan, the study reconstructed abdominal wall structures in 3D and analyzed how successive surgical steps redistributed stress across fascial tissues. The simulations reproduced clinically observed tension-reduction trends, particularly within the posterior rectus sheath, while also identifying model assumptions that may influence prediction accuracy. Biomechanics-informed surgical modeling could help tailor hernia repair strategies, reduce recurrence risk, and support long-term recovery after complex abdominal wall reconstruction. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Patient-Specific Computational Model of Abdominal Wall Simulating Hernia Defect Closure to Predict Tension in the Posterior and Anterior Rectus Sheath 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Manar Jellal, Arthur Jourdan, Gaëtan-Romain Joliat, Guillaume Passot, and Anicet Le Ruyet @unil @medtronic 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Accurately recreating the complex environment inside a living tooth remains difficult, slowing efforts to improve dental therapies and study how teeth respond to damage. Examining progress in this area, researchers reviewed the development and application of microfluidic devices designed to mimic the dentin-pulp complex. Across five in vitro studies, the analysis found wide variation in device design, fabrication methods, and reporting standards, with most models relying on static cultures rather than dynamic fluid flow that closely reflects real biological conditions. The authors also highlighted inconsistent characterization of device features, limiting reproducibility and comparison across studies. Improved physiologic dentin-pulp models could sharpen dental research, accelerate biomaterial testing, and bring laboratory findings closer to real-world clinical performance. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗥𝗲𝘃𝗶𝗲𝘄: Organ-on-a-Chip Devices to Simulate the Dentin-Pulp Complex: A Qualitative Systematic Review 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Josefa Baeza-Fernández and @CristinaBucchi1 @UFrontera 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Bone-anchored prostheses are helping more people with below-knee amputations regain mobility, but long-term implant durability and infection risk remain critical challenges. Researchers reviewed the evolution of bone-anchored prosthesis treatment for transtibial amputation using both published evidence and 10 years of clinical experience involving 124 individuals and 133 osseointegration implants. Early procedures were associated with high rates of infection and implant loosening, but subsequent advances in implant design and surgical technique led to improved functional outcomes and relatively low rates of major complications. The review also highlighted unresolved questions surrounding implant placement, fixation strategies, patient selection, and the amount of bone integration needed to support long-term success. As osseointegration systems continue to mature, improving implant longevity and reliability could expand access to prosthetic solutions that feel more stable, natural, and durable in everyday life. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗥𝗲𝘃𝗶𝗲𝘄: Tibial Bone-Anchored Prostheses: A Narrative Review of the Literature and Reflection on 10 Years of Surgical Practice 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Robin Atallah, Jan Paul Frölke, and Ruud Leijendekkers @maartenskliniek @radboudumc 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Learning to use a powered exoskeleton effectively on the job may take far more training than simply putting the device on and getting started. Engineers compared novice and experienced users performing simulated occupational load-handling tasks while wearing a whole-body powered exoskeleton. Across three training sessions, novice users cut task completion time nearly in half and reduced movement jerk by 30%, while also showing improvements in movement patterns, muscle activation, and perceived workload. Even so, important differences remained – including slower movement speeds, greater hip flexion, and reduced coordination efficiency compared to experienced users – suggesting that mastering these systems requires extended familiarization beyond a few sessions. Improved training protocols could be just as important as more advanced exoskeletons in helping reduce workplace strain, fatigue, and injury risk. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Adaptation to a Whole-Body Powered Exoskeleton: Human–Exoskeleton Coordination During Load-Handling Tasks 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Hanjun Park, Sunwook Kim, Maury A. Nussbaum, and Divya Srinivasan @TexasTech @TTUEngineering @Ttuimse @virginia_tech @vtengineering @ICTASVT @ClemsonUniv @ClemsonCECAS @ClemsonIE 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Swelling and fluid buildup can become serious problems when the body’s lymphatic drainage system does not work as effectively as it should. To clarify what drives lymphatic fluid transport, researchers developed a computational model that simulated how lymphatic vessels respond to active pumping, external mechanical forces, and surrounding tissue properties across different branching architectures. The simulations showed that highly branched lymphatic networks produced lower flow overall and relied more heavily on active vessel pumping, while interstitial permeability strongly controlled passive drainage. Sensitivity analyses further revealed that vessel compliance and permeability interact differently depending on network complexity, highlighting how local tissue structure can reshape lymphatic function. A deeper understanding of how lymphatic networks manage fluid transport could help guide more targeted approaches for conditions linked to impaired drainage and chronic swelling. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: A Modular Computational Approach for Assessing Active and Passive Force Contributions in Interstitial Lymphatic Fluid Uptake 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Tharanga D. Jayathungage Don, Finbar Argus, @soroushsaf, Peter S. Russell, Anthony RJ. Phillips, and Hayley M. Reynolds @AucklandUni @ABI_bioeng @FMHS_UoA @STaR_Centre_NZ 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Mechanical heart pumps used during cardiac emergencies must balance strong blood flow support with minimizing damage to blood cells – but designing those systems efficiently has remained a major engineering hurdle. Researchers aimed to improve blood flow support while reducing motor demands by developing an optimization framework for percutaneous ventricular assist devices that tuned the motor and impeller together rather than separately. The method combined artificial neural networks, analytical target cascading, and a multi-objective optimization algorithm to reduce computational burden while improving performance tradeoffs. The resulting prototype generated clinically relevant pressure and flow levels, maintained a low hemolysis index, and improved overall efficiency from 7.59% in the baseline design to 33.65%. Smarter optimization strategies could accelerate development of safer, higher-performing circulatory support devices when every second of cardiac support matters. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Multidisciplinary Optimization Design of pVADs Using Analytical Target Cascading-Guided Genetic Algorithm 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Chenghan Chen, Yuyang Shi, E. Chaoran, Hao Qian, Pengfei Hao, Feng He, Mingkui Zhang, and Xiwen Zhang @Tsinghua_Uni 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Constipation can be difficult to treat with abdominal massage because clinicians still lack clear, personalized standards for how much pressure massage devices should apply. Seeking to establish quantitative standards for abdominal massage, researchers combined finite element modeling and machine learning to create a digital twin of the abdomen using CT images and biomechanical data. They developed a high-fidelity skin–muscle model, then trained stacked learning models to predict how massage force and abdominal wall deformation interact across different regions of the colon. The system achieved prediction accuracies exceeding 84% for massage force and up to 96% for abdominal wall deformation while generating results in just 3 seconds compared with traditional simulation workflows. Individualized and quantitatively guided abdominal therapies could improve symptom relief while making noninvasive digestive care faster, safer, and more consistent. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Prediction of Massage Force and Intra-abdominal Wall Deformation During Massage by a Digital Twin Model Based on an Abdominal Finite Element Model 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Junjie Fu, Yanchen Du, Xinyi Tang, and Hongliu Yu 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Many strokes linked to atrial fibrillation are driven by hidden blood flow patterns that current risk scores don’t capture. Seeking to bridge this gap, researchers used high-fidelity computational fluid dynamics to analyze blood flow in 40 patient-specific left atrium models reconstructed from CT. They quantified flow and structural features and tested how these metrics relate to thrombus presence or stroke history. Several parameters – including left atrial appendage geometry and blood residence time – showed strong associations, and when combined with the CHA₂DS₂–VASc score, improved predictive performance. Better risk stratification could enable more precise decisions about monitoring and treatment, reducing avoidable stroke events. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Beyond CHA₂DS₂–VASc: Hemodynamic and Morphologic Discriminants for Thrombus Formation and Stroke in Atrial Fibrillation Patients 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Henrik Aasen Kjeldsberg, Josquin Harrison, @MaximeSermesant, Renate B. Schnabel, Joakim Sundnes, and Kristian Valen-Sendstad @simula_research @Inria @dzhk_germany 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

When life support systems meant to save lives also trigger dangerous bleeding, how do we make them safer? Researchers tested whether a redesigned extracorporeal membrane oxygenation system, the InFlo MOBYBOX ECMO system, could maintain clot control under high-risk, low-flow conditions without relying on systemic anticoagulation – a key driver of bleeding complications. In a seven-day preclinical study under low-flow conditions and without systemic anticoagulants, the system maintained stable oxygen and carbon dioxide exchange while showing minimal clot formation, no hemolysis, and no organ damage. These findings point to a multimodal design that mitigates both flow-driven and surface-mediated contributors to thrombosis. Reducing dependence on blood thinners during extracorporeal support could lower bleeding risk while preserving life-sustaining function. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Prolonged Extracorporeal Membrane Oxygenation Without Systemic Anticoagulation in Sheep 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Zahra Mahdieh, Alex Meng, Brian Y. Chang, and Nicholas X. Williams 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Understanding how bone behaves across scales remains a major barrier to improving how we predict fractures and design effective treatments. Reviewing recent advances in this area, researchers compared classical computational models with data-driven approaches for capturing the multiscale structure and mechanics of bone. They evaluated how each framework handles complex behavior, incorporates biological data, and balances computational cost with interpretability, while also highlighting hybrid methods that combine their strengths. To translate these insights into practice, they proposed a hierarchical decision matrix to guide model selection based on factors such as data availability, modeling goals, and computational constraints. Smarter model selection and integration across scales could sharpen fracture risk prediction, improve treatment planning, and support personalized care. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗥𝗲𝘃𝗶𝗲𝘄: Capturing the Multiscale Nature of Bone Behavior: Classical, Data-Driven and Hybrid Techniques 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Melika Mohammadkhah, Ardeshir Savari, and Sandra Klinge 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Why do people with Parkinson’s disease have difficulty adjusting their decisions when reward patterns suddenly change? Researchers examined the drivers of this behavior by applying a neurocomputational model to simulate how 18 individuals with Parkinson’s disease and 14 control subjects perform in a probabilistic reversal learning task. By fitting parameters tied to dopamine levels, learning dynamics, and exploratory behavior, the model reproduced individual performance across both stable and changing reward conditions. The analysis showed that differences in dopamine levels tracked medication state, while learning rate and exploratory tendencies explained how people respond to punishment and adjust choices. Characterizing how these distinct mechanisms shape adaptive behavior could support more effective, customized approaches to managing cognitive flexibility in neurological conditions. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: A Mechanistic Understanding of the Different Factors Affecting Adaptation and Flexibility During Probabilistic Reversal Learning: A Neurocomputational Study on Patients with Parkinson’s Disease and Control Subjects 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Giulia Piermaria, Silvana Pelle, Alessio Zizzi, Carola Cosentino, Susanna Mezzarobba, Giovanna Calandra-Buonaura, Luisa Sambati, Federica Scarpellini, Elisa Pelosin, and Mauro Ursino @unibo 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Subtle changes in tissue structure often go undetected, limiting how well we can understand and treat disease and healing. To improve how these differences are identified, researchers evaluated a statistical approach called Statistical nonParametric Mapping using both simulated and experimental datasets of collagen fibril diameters. Traditional methods could detect broad overall changes but often missed subtler shifts or produced results that depended heavily on how the data were grouped. In contrast, this approach pinpointed where specific differences occurred within the diameter distribution, providing a more detailed and reliable picture of tissue remodeling. More sensitive detection of microstructural changes could translate into earlier diagnosis, better monitoring of healing, and more informed therapeutic decisions. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Statistical nonParametric Mapping Enables Rigorous Comparison of Collagen Fibril Diameter Distributions 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Jeremy D. Eekhoff and Louis J. Soslowsky @Penn @PennMedicine @McKayLaboratory @SoslowskyLab 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Shoulder injuries are hard to diagnose and treat because the body can compensate for weak muscles in ways that mask the root problem. Using a detailed computational model, researchers applied optimal control methods to simulate how the shoulder adapts to isolated muscle weakness during arm elevation. Key findings from the simulations include: 🔹 Weakness in the serratus anterior produced the largest changes in shoulder blade motion and triggered the most compensation from other muscles, with model predictions aligning well with experimental data 🔹 Weakness in the upper trapezius resulted in minimal compensatory muscle activity 🔹 Movement in the scapular plane reduced coactivation of trapezius muscles, while frontal plane motion promoted more balanced engagement By uncovering how specific muscles drive compensation, this approach could sharpen diagnosis and guide more targeted rehabilitation strategies for shoulder dysfunction. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Shoulder Kinematic and Muscle Activity Compensations to Scapular Stabilizer Weakness: An Optimal Control Framework 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Matthew S. Russell, Daanish M. Mulla, @KeirPeter, Edward K. Chadwick, @DimitraBlana, Janessa D. M. Drake, and Jaclyn N. Chopp-Hurley @YorkUniversity @YorkUKINE @McMasterU @MacKinesiology @aberdeenuni @AbdnLifeScience 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Predicting which drugs may trigger dangerous heart rhythm disturbances is an ongoing challenge in preclinical safety assessment. Working to clear this hurdle, researchers built a statistical model that combines signals from different heart cell types rather than relying on just one. They analyzed ion channel effects for 28 compounds, calculated a risk-related metric across key ventricular cell layers, and then averaged the model’s predictions to produce a single, more stable result. Tested on established datasets, the approach consistently outperformed standard methods and delivered near-perfect classification performance. More reliable early screening of serious heart rhythm side effects could help ensure safer therapies reach the clinic with greater confidence. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: Improving In Silico Cardiac Safety Prediction by Consensus Averaging of Transmural Ventricular Cell Models 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Nurul Qashri Mahardika, Ali Ikhsanul Qauli, Yunendah Nur Fuadah, Aulia Khamas Heikhmakhtiar, Muhammad Adnan Pramudito, Ariyadi, Aroli Marcellinus, Yoo Seok Kim, and Ki Moo Lim @Unair_Official @TelUniversity @Unhan_RI 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Clinicians and engineers need precise ways to determine how soft tissues behave under load, but extracting these properties from real-world data remains difficult. Researchers sought a solution by developing a computational framework that estimates tissue stiffness, nonlinear behavior, and fiber orientation directly from how tissues deform under force. Instead of repeatedly running full simulations, the method works backward from measured deformation patterns to recover material properties more efficiently, while a multilayer perceptron captures how those properties vary across complex structures. Tests ranging from synthetic tissue cases to a bioprosthetic heart valve showed faster optimization than Physics-Informed Neural Network approaches and consistently recovered fiber directions even in challenging 3D geometries. More detailed identification of tissue mechanics could lead to better-targeted interventions, improved device performance, and more predictable outcomes in complex biomechanical applications. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: A Gradient-Based Machine-Learning Inspired Inverse Modeling Approach for Characterization of Nonlinear Tissue Properties 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Hossein Geshani and Iman Borazjani @TAMU @TAMUEngineering 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439

Noninvasive heart scans can estimate whether coronary blockages restrict blood flow, but key assumptions about the smallest vessels still limit their accuracy. Researchers addressed this by developing a method to estimate how the heart’s smallest vessels respond under stress using routinely collected patient data, then incorporating these insights into CT-based fractional flow reserve calculations. They derived a statistical model linking microvascular flow response to factors such as sex, diabetes and smoking status using perfusion data, then applied it to patient-specific simulations and validated results against invasive fractional flow reserve measurements. The approach improved accuracy over conventional methods and produced values closely aligned with invasive measurements, without requiring additional perfusion imaging. More reliable and accessible noninvasive assessments could sharpen clinical decision-making and reduce unnecessary procedures in cardiovascular care. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10.100… 𝗦𝘁𝘂𝗱𝘆: A Novel Demography-Based Approach to Define Patient-Specific Outflow Boundary Conditions in CT-Based FFR Computations 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Ernest W. C. Lo, @DrFranPugliese, Leon Menezes, and Ryo Torii @UCL @UCLEngineering @UCLmedphys @uclmecheng @medimaging_cdt @UCLHresearch @QMUL @QMULWHRI 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10439