Conference Agenda

Investigating intracardiac flow dynamics is crucial for diagnosing and treating congenital heart defects, as flow patterns can indicate disease progression and inform the timing of therapeutic interventions. While 4D flow MRI enables non-invasive quantification and visualization of the cardiovascular flow field, its clinical applicability is constrained by limited spatial and temporal resolution, which restricts its capability to resolve fine-scale flow features and derived hemodynamic parameters such as vorticity, wall shear stress, and energy loss. These limitations are especially important in pediatric cases, where small anatomical scales and rapid heart rates requires higher resolution. Computational tools can fill this gap and while significant progress has been made in developing advanced multi-physics computational models that account for the complex interactions of elasto-mechanics, electrophysiology and hemodynamics in the heart, their direct application to patient specific cases in congenital heart disease remains problematic. This is due to their dependance the constitutive relations used for example to model fluid-solid interactions that rely heavily on experimentally derived material properties, which may not accurately represent patient-specific or pathological conditions.

In this work we propose a cost-effective alternative that reduces model complexity by replacing multiphysics components with prescribed kinematics derived from patient-specific imaging data. This approach preserves flow-related information and resolution while drastically reducing computational costs, making it more suitable for clinical applications. It also facilitates model adaptation to both healthy and pathological cases, offering advantages in pediatric cardiology, where rapid, patient-specific analysis is essential. Despite these benefits, reconstructing patient-specific geometries—particularly for thin, dynamic structures like the tricuspid valve remains a significant challenge. The delicate leaflets are often poorly resolved in conventional imaging modalities due to temporal and spatial resolution constraints. Although previous studies have proposed structural models of the tricuspid valve with varying complexity, most rely on experimental material properties that are difficult to individualize, limiting their applicability in patient-specific scenarios. Here we present a novel, computationally efficient model derived entirely from 4D-MRI velocity data. Our approach is based on the observation that the gradient of the velocity magnitude is highest near vascular boundaries, including the tricuspid valve leaflets, due to impermeability. Segmentation and registration of the right ventricle (RV) and right atrium (RA) were performed using 3D Slicer, an open-source platform for medical image processing. The final kinematic reconstruction of the RV, including the valve geometry, was prescribed using the Large Deformation Diffeomorphic Metric Mapping (LDDMM) framework.

We conducted direct numerical simulations (DNS) using an in-house solver based on the Immersed Boundary Method (IBM) to solve the Navier–Stokes equations for incompressible flows. The 4D-MRI velocity field served as a benchmark, and DNS results were compared both qualitatively (velocity field during diastolic filling) and quantitatively. Our findings demonstrated strong agreement between DNS and 4D-MRI, particularly in capturing diastolic RV filling patterns in patient-specific cases. This validates our model’s ability to replicate physiological flow without relying on expensive multi-physics simulations, offering a scalable tool for clinical hemodynamic assessment.

To tackle some of the current challenges in robotics, such as adaptability in complex and unpredictable environments, as well as miniaturization constrains, scientists have engineered bio-bots, by integrating living cells with synthetic components. Beyond advancing robotics, these biohybrid platforms offer valuable insights into the fundamental biophysics and biology of natural systems, with potential applications in human health and medicine. Despite their promising capabilities, significant challenges remain, including ethical implications, technical roadblocks in control strategies and the development of predictive computational modelling tools to accurately design, simulate and optimize the biohybrid prototypes.

Taking inspiration from the biophysics of the human heart, we have conceived a new class of biohybrid swimmers. The envisioned system consists of a hollow cylinder made of cardiac muscle cells (cardiomyocytes), integrated with an artificial activation device. As in the human heart, the sinoatrial node triggers the synchronized contraction of muscle fibres to enable rhythmic blood pumping, the propulsion mechanism of our prototype is driven by an electric current generated by a pacemaker situated on its leading edge. This impulse trigger periodic muscular contractions, resulting in the propagation of peristaltic elastic waves along the soft body of the swimmer, providing the thrust for the forward locomotion, in the low to intermediate Reynolds number regime.

More specifically, the motion of the swimmer results from the interconnected dynamics of the electrophysiology system, responsible for the action potential propagation that triggers the active muscular tension, the internal passive forces of the biological tissue, and the hydrodynamic loads due to the interaction with the surrounding fluid. By simultaneously accounting for the electrophysiology, elastomechanics, and hydrodynamics, as well as their interaction, our state-of-the-art computational tool allows us to optimize and predict the swimming dynamics through three-dimensional numerical simulations.

In principle, the swimming characteristics, optimal configuration, excitation frequency and physical specifications of the system are highly dependent on the specific condition being simulated. However, our findings demonstrate that the system exhibits unexpectedly robust features. Each configuration analysed is characterized by an optimal excitation frequency, maximizing both swimming speed and initial acceleration. Notably, the optimal actuation frequency is Reynolds-independent, indicating that viscosity does not significantly impact the propulsion mechanism and is therefore not a key parameter in the design of such structures.

To elucidate the behaviour of the system, we propose a physics-based analytical model capable of accurately predicting the swimming dynamics in the optimal scenario. Interestingly, our findings show that peak performance is achieved when a single peristaltic wave propagates along the swimmer body, whereas existing models in the Stokes regime predict that increasing the number of pulses enhances swimming speed. In particular, once the contraction reaches the distal end, the peristaltic cycle immediately resumes, providing uninterrupted propulsion. Finally, we capture the fundamental aspects of the system’s locomotion strategy by deriving a universal scaling that links the swimming speed to body kinematics.

In conclusion, the proposed proof-of concept provides useful design guidelines for the development of future biohybrid devices, which can lead to new insights in robotics, bioengineering and medicine.

Nowadays, numerical simulations are valuable tools for cardiovascular research. They allow for testing new devices and surgical interventions and evaluating quantities difficult to measure in vivo or in vitro. Nevertheless, the reliability of their results is strongly linked to the assumptions and uncertainties in the computational model. The selection of an appropriate constitutive relation for blood viscosity represents a crucial concern for computational models of haematic flows.
Blood is a biphasic fluid consisting of a solid phase, which includes red blood cells (RBCs), white blood cells (WBCs) and platelets, and a liquid phase, i.e. plasma. Viscoelastic, shear-thinning, and thixotropic properties of blood are mainly attributed to the dynamics and interactions of RBCs. Therefore, non-Newtonian rheology should be taken into account when modelling blood flows. However, when strains in the flow are higher than a given threshold, considering blood as a Newtonian fluid is a reasonable hypothesis. For this reason, several previous studies adopted this assumption, especially for flows in large vessels. Nonetheless, it may not be adequate under pathological conditions and at instants of the cardiac cycle when recirculation regions are present. Previous works compared Newtonian and non-Newtonian rheological models for flows through heart valves, arteries, and single cardiac chambers, often introducing some simplifications. In the present study, we perform numerical simulations of the fluid-structure-electrophysiology interaction (FSEI) in the left heart with an in-house developed, GPU-accelerated code. We assume the Newtonian and Carreau constitutive law for blood viscosity. We assess the influence of the rheological model on haemodynamic quantities of interest for clinical practice. Specifically, non-Newtonian rheology does not substantially affect physical quantities depending on the volumetric stress tensor, such as the pressure in the chambers and aorta or the orifice area of the valves. On the other hand, we observe considerable differences in the wall shear stress and local haemolysis between the Newtonian and non-Newtonian models. Setting the constant viscosity in the Newtonian model equal to the spatio-temporal average of the value estimated by the Carreau model does not mitigate the discrepancies. For this reason, non-Newtonian rheological models are recommended if interested in red blood damage, tissue remodelling and atherogenesis.
Future work aims to include more complex non-Newtonian properties in the rheological model, such as thixotropicity and viscoelasticity. Moreover, since previous studies highlighted that non-Newtonian effects are more evident when diseases are present, the analysis will be repeated for a left heart under pathological conditions. Additionally, the effect of haematocrit will also be investigated with the goal of improving the accuracy of patient-specific simulations.

The contraction of the heart during a heartbeat is the consequence of the synchronized reaction of the myocardium to the propagation of a non-linear electrophysiology wave. Such scenario, however, can be altered in pathological conditions, such as when ventricular tachycardia or ventricular fibrillation manifest. These phenomena are related to the non physiological tissue activation under the propagation of non-planar electrical waves, namely spiral and scroll waves, which are often triggered by the presence of defects introduced by myocardial infarction. In particular, the scar region is known to act as a support for the generation of scroll waves owing to the high conductivity difference between healthy and pathological tissue. While the dynamics and formation of arrhytmogenic patterns is widely studied in literature [ 1, 2], their consequences on the cardiac hemodynamics are rarely taken into account. In this work, we aim at investigating what are the effects of ventricular tachycardia on the more relevant bio-markers, in particular, determine what pathological electrophysiology patterns have the most significant impact on the physiological blood flow dynamics in the left heart and try to understand the optimal treatment. To this aim, the tissue activation data of a patient affected by an arrhythmogenic ventricular scar mapped through the CARTO^® 3 system, has been used in order to fine tune the heterogeneities of the substrate for the electrophysiology model in order to trigger reentries in the myocardium. The electrophysiology is then solved trough a monodomain model coupled with suitable cellular models encompassing the hetoregeneity of the myocardium owing to the presence of
healthy tissue, peri-infarct region and scar tissue. The corresponding hemodynamics is then solved by integrating the Navier-Stokes equations, which are discretized using a staggered finite differences complemented with immersed boundary techinques [3]. The resulting multi-physics system can be then exploited as a predictive tool to reproduce ventricular arrhythmias and try to understand what is the optimal intervention in order to restore the physiological cardiac function.

¹ Salvador M. et al., Computers in Biology and Medicine, 142, 2022.

² Ramírez W. A. et al., Scientific Reports, 10, 2020.

³ Viola F. et al., Scientific Reports, 13, 2023.