Explore the vast range of titles available.

Three-dimensional (3D) printing is emerging as a transformative technology for biomedical engineering. The 3D printed product can be patient-specific by allowing customizability and direct control of the architecture. The trial-and-error approach currently used for developing the composition of printable inks is time- and resource-consuming due to the increasing number of variables requiring expert knowledge. Artificial intelligence has the potential to reshape the ink development process by forming a predictive model for printability from experimental data. In this paper, we constructed machine learning (ML) algorithms including decision tree, random forest (RF), and deep learning (DL) to predict the printability of biomaterials. A total of 210 formulations including 16 different bioactive and smart materials and 4 solvents were 3D printed, and their printability was assessed. All ML methods were able to learn and predict the printability of a variety of inks based on their biomaterial formulations. In particular, the RF algorithm has achieved the highest accuracy (88.1%), precision (90.6%), and F1 score (87.0%), indicating the best overall performance out of the 3 algorithms, while DL has the highest recall (87.3%). Furthermore, the ML algorithms have predicted the printability window of biomaterials to guide the ink development. The printability map generated with DL has finer granularity than other algorithms. ML has proven to be an effective and novel strategy for developing biomaterial formulations with desired 3D printability for biomedical engineering applications.

Red blood cells (RBCs) are a key determinant of human physiology and their behaviour becomes extremely heterogeneous as they navigate in narrow, bifurcating vessels in the microvasculature, affecting local haemodynamics. This is due to partitioning in bifurcations which is dependent on the biomechanical properties of RBCs, especially deformability. We examine the effect of deformability on the haematocrit distributions of dense RBC suspensions flowing in a single, asymmetric Y-shaped bifurcation, experimentally. Human RBC suspensions (healthy and artificially hardened) at 20% haematocrit (Ht) were perfused through the microchannels at different flow ratios between the outlet branches, and negligible inertia, and imaged to infer cell distributions. Notable differences in the shape of the haematocrit distributions were observed between healthy and hardened RBCs near the bifurcation apex. These lead to more asymmetric distributions for healthy RBCs in the daughter and outlet branches with cells accumulating near the inner channel walls, exhibiting distinct hematocrit peaks which are sharper for healthy RBCs. Although the hematocrit distributions differed locally, similar partitioning characteristics were observed for both suspensions. Comparisons with RBC distributions measured in a T-shaped bifurcation showed that the bifurcation angle affects the haematocrit characteristics of the healthy RBCs and not the hardened ones. The extent of RBC partitioning was found similar in both geometries and suspensions. The study highlights the differences between local and global characteristics which impact RBC distribution in more complex, multi-bifurcation networks.

The optimal treatment of Type-B aortic dissection (AD) is still a subject of debate, with up to 50% of the cases developing late-term complications requiring invasive intervention. A better understanding of the patient-specific haemodynamic features of AD can provide useful insights on disease progression and support clinical management. In this work, a novel in vitro and in silico framework to perform personalised studies of AD, informed by non-invasive clinical data, is presented. A Type-B AD was investigated in silico using computational fluid dynamics (CFD) and in vitro by means of a state-of-the-art mock circulatory loop and particle image velocimetry (PIV). Both models not only reproduced the anatomical features of the patient, but also imposed physiologically-accurate and personalised boundary conditions. Experimental flow rate and pressure waveforms, as well as detailed velocity fields acquired via PIV, are extensively compared against numerical predictions at different locations in the aorta, showing excellent agreement. This work demonstrates how experimental and numerical tools can be developed in synergy to accurately reproduce patient-specific AD blood flow. The combined platform presented herein constitutes a powerful tool for advanced haemodynamic studies for a range of vascular conditions, allowing not only the validation of CFD models, but also clinical decision support, surgical planning as well as medical device innovation.

Although bulk changes in red blood cell concentration between vessels have been well characterised, local distributions are generally overlooked. Red blood cells aggregate, deform and migrate within vessels, forming heterogeneous distributions which have considerable effect on local haemodynamics. The present study reports data on the local distribution of human red blood cells in a sequentially bifurcating microchannel, representing the branching geometry of the microvasculature. Imaging methodologies with simple extrapolations are used to infer three dimensional, time-averaged velocity and haematocrit distributions under a range of flow conditions. Strong correlation between the bluntness of the velocity and haematocrit profiles in the parent branch of the geometry is observed and red blood cell aggregation has a notable effect on the observed trends. The two branches of the first bifurcation show similar characteristics in terms of the shapes of the profiles and the extent of plasma skimming, despite the difference in geometric configuration. In the second bifurcation, considerable asymmetry between the branches in the plasma skimming relationship is observed, and elucidated by considering individual haematocrit profiles. The results of the study highlight the importance of considering local haematocrit distributions in the analysis of blood flow and could lead to more accurate computational models of blood flow in microvascular networks. The experimental approaches developed in this work provide a foundation for further examining the characteristics of microhaemodynamics.

Background The management and prognosis of aortic dissection (AD) is often challenging and the use of personalised computational models is being explored as a tool to improve clinical outcome. Including vessel wall motion in such simulations can provide more realistic and potentially accurate results, but requires significant additional computational resources, as well as expertise. With clinical translation as the final aim, trade-offs between complexity, speed and accuracy are inevitable. The present study explores whether modelling wall motion is worth the additional expense in the case of AD, by carrying out fluid-structure interaction (FSI) simulations based on a sample patient case. Methods Patient-specific anatomical details were extracted from computed tomography images to provide the fluid domain, from which the vessel wall was extrapolated. Two-way fluid-structure interaction simulations were performed, with coupled Windkessel boundary conditions and hyperelastic wall properties. The blood was modelled using the Carreau-Yasuda viscosity model and turbulence was accounted for via a shear stress transport model. A simulation without wall motion (rigid wall) was carried out for comparison purposes. Results The displacement of the vessel wall was comparable to reports from imaging studies in terms of intimal flap motion and contraction of the true lumen. Analysis of the haemodynamics around the proximal and distal false lumen in the FSI model showed complex flow structures caused by the expansion and contraction of the vessel wall. These flow patterns led to significantly different predictions of wall shear stress, particularly its oscillatory component, which were not captured by the rigid wall model. Conclusions Through comparison with imaging data, the results of the present study indicate that the fluid-structure interaction methodology employed herein is appropriate for simulations of aortic dissection. Regions of high wall shear stress were not significantly altered by the wall motion, however, certain collocated regions of low and oscillatory wall shear stress which may be critical for disease progression were only identified in the FSI simulation. We conclude that, if patient-tailored simulations of aortic dissection are to be used as an interventional planning tool, then the additional complexity, expertise and computational expense required to model wall motion is indeed justified.

Modelling blood flow in microvascular networks is challenging due to the complex nature of haemorheology. Zero- and one-dimensional approaches cannot reproduce local haemodynamics, and models that consider individual red blood cells (RBCs) are prohibitively computationally expensive. Continuum approaches could provide an efficient solution, but dependence on a large parameter space and scarcity of experimental data for validation has limited their application. We describe a method to assimilate experimental RBC velocity and concentration data into a continuum numerical modelling framework. Imaging data of RBCs were acquired in a sequentially bifurcating microchannel for various flow conditions. RBC concentration distributions were evaluated and mapped into computational fluid dynamics simulations with rheology prescribed by the Quemada model. Predicted velocities were compared to particle image velocimetry data. A subset of cases was used for parameter optimisation, and the resulting model was applied to a wider data set to evaluate model efficacy. The pre-optimised model reduced errors in predicted velocity by 60% compared to assuming a Newtonian fluid, and optimisation further reduced errors by 40%. Asymmetry of RBC velocity and concentration profiles was demonstrated to play a critical role. Excluding asymmetry in the RBC concentration doubled the error, but excluding spatial distributions of shear rate had little effect. This study demonstrates that a continuum model with optimised rheological parameters can reproduce measured velocity if RBC concentration distributions are known a priori. Developing this approach for RBC transport with more network configurations has the potential to provide an efficient approach for modelling network-scale haemodynamics.

High-fidelity numerical simulations such as Computational Fluid Dynamics (CFD) have been proven effective in analysing haemodynamics, offering insight into many vascular conditions. However, these methods often face challenges of high computational cost and long processing times. Data-driven approaches such as Reduced Order Modeling (ROM) and Machine Learning (ML) are increasingly being explored alongside CFD to advance biomechanical research and application. This study presents an integration of Proper Orthogonal Decomposition (POD)-based ROM with neural network-based ML models to predict Wall Shear Stress (WSS) in patient-specific vascular pathologies. CFD was used to generate WSS data, followed by POD to construct the ROM. The ML models were trained to predict the ROM coefficients from the inlet flowrate waveform, which can be routinely collected in the clinic. Two ML models were explored: a simpler flowrate-coefficients mapping model and a more advanced autoregressive model. Both models were tested against two case studies: flow in Peripheral Arterial Disease (PAD) and flow in Aortic Dissection (AD). Despite the limited training data sets (three flowrate waveforms for the PAD case and two for the AD case), the models were able to predict the haemodynamic indices, with the flowrate-coefficients mapping model outperforming the autoregressive model in both case studies. The accuracy is higher in the PAD case study, with reduced accuracy in the more complex case study of AD. Additionally, the computational cost analysis reveals a significant reduction in computational demands, with speed-up ratios in the order of 10 4 for both case studies. This approach shows an effective integration of ROM and ML techniques for fast and reliable evaluations of haemodynamic properties that contribute to vascular conditions, setting the stage for clinical translation.

Microscale blood flow is characterised by heterogeneous distributions of hematocrit, viscosity and velocity. In microvascular bifurcations, cells are unevenly distributed between the branches, and this effect can be amplified in subsequent branches depending on a number of parameters. We propose an approach to infer hematocrit profiles of human blood flowing through a bifurcating microchannel. The influence of aggregation, induced by the addition of Dextran 2000 to the samples, is also considered. Averaged values indicate plasma skimming, particularly in the presence of red blood cell (RBC) aggregation. Using an empirical model, the hematocrit profiles are used to estimate local relative viscosity distributions. Simulations are used to predict how the non-uniform viscosity influences the velocity profiles. Comparing these data to velocity profiles of RBCs measured using particle image velocimetry provides validation of the model. It is observed that aggregation blunts velocity profiles after a long straight section of channel. Downstream of the bifurcation, skewing of the velocity profiles is detected, which is enhanced by aggregation. The proposed methodology is capable of providing hitherto unreported information on important aspects of microscale blood rheology.

We study the shear-thinning mediation of elasto-inertial transitions in Taylor–Couette flow of viscoelastic polymer solutions. Two types of high molecular weight polymers are used at various concentrations and in water–glycerol solvents of various viscosities. This allows us to access a wide range of elastic numbers and effective shear-thinning indices. Conservative ramp-up (slow acceleration of the inner cylinder and subsequent increase in Reynolds number) and steady-state (constant rotation speed) experiments are performed, in which the flow is monitored continuously using flow visualisation. Depending on the shear-thinning and elastic properties of the working fluid, very different behaviours are observed. In almost constant-viscosity fluids (Boger fluids), or shear-thinning fluids with significant elasticity, the flow transitions from purely azimuthal Couette flow (CF) to a highly chaotic flow state referred to as elasto-inertial turbulence (EIT) via Taylor vortex flow (TVF) and elasto-inertial rotating spiral waves (RSW). When the degree of shear-thinning is increased and elasticity reduced, elastic waves or EIT may fade to a wavy Taylor vortex flow (WTVF) with increasing inertia. Significant shear-thinning leads to a delay in the onset of EIT. Remarkably, in some highly shear-thinning cases, even with a significant elasticity, elastic flow features (EIT, RSW) are completely suppressed, and the flow exhibits a ‘Newtonian-like’ transition sequence (CF–TVF–WTVF). Shear-thinning acts to modify, delay, or even completely suppress elasto-inertial behaviours (RSW, EIT), that would otherwise have existed in the absence of shear-thinning. It is, thus, possible to induce various hydrodynamic regimes by tuning the relative degrees of shear-thinning, elasticity and inertia.

We study Taylor–Couette flow of a glycerol–water mixture containing a wide range of concentration (0–2000 ppm) of xanthan gum, which induces both shear-thinning and viscoelasticity, in order to assess the effect of the changes in rheology on various flow instabilities. For each suspension, the Reynolds number (the ratio of inertial to viscous forces) is slowly increased to a peak value of around 1000, and the flow is monitored continuously using flow visualisation. Shear-thinning is found to suppress many elasticity-controlled instabilities that have been observed in previous studies of viscoelastic Taylor–Couette flow, such as diwhirls and disordered oscillations. The addition of polymers is found to reduce the critical Reynolds number for the formation of Taylor vortices, but delay the onset of wavy flow. However, in the viscoelastic regime (${\\geq }1000\\ \\textrm {ppm}$ concentration), the flow becomes highly unsteady soon after the formation of Taylor vortices, with substantial changes in the waviness with Reynolds number, which are shown to be highly repeatable. Vortices are found to suddenly merge as the Reynolds number increases, with the number of mergers increasing with polymer concentration. These abrupt changes in wavelength are highly hysteretic and can occur in both steady and wavy regimes. Finally, the vortices in moderate and dense polymer solutions are shown to undergo a gradual drift in both their size and position, which appears to be closely linked to the splitting and merger of vortices.