Robotic right ventricle is a biohybrid platform that simulates right ventricular function in (patho)physiological conditions and intervention

Figures

• 

  Fig. 1. An overview of the RRV and its architecture. a, Overview of the bioinspired and biomimetic approach used to create a biohybrid beating right heart. Initially, a freshly explanted porcine heart underwent chemical treatment, after which the native myocardium was hand-dissected and replaced with a robotic counterpart while preserving the endocardial scaffold. b, The assembly of the RRV was imaged with micro-CT, revealing the preserved intracardiac structures. c, Schematic of the RV’s complex shape, fiber orientation and wall motion (left). The image of the heart showing fiber orientation in the basal and apical loops is reprinted from Buckberg et al.12, with permission from Elsevier. The physical model of the soft robotic myocardium shows the placement of individual actuators in the synthetic myocardium (right). The outflow tract of the RV contains infundibular muscles that also contribute to RV ejection through contraction. The contraction of the outflow region can be achieved by placing a circumferential actuator across the infundibulum. The numbers denote the respective actuator placement corresponding to the RV motion shown in the schematics on the left. SVC, superior vena cava.

• 

  Fig. 2. Computational and in vitro modeling of the RV biomechanics. a, FE model of the soft robotic myocardium, highlighting the placement and spatial density of the circumferential and longitudinal actuators. The computational myocardium exhibits structural deformations during both diastolic and systolic phases, which can be observed from long-axis (top) and short-axis (bottom) views. b, Schematics illustrating the design principles of modeling two-way fluid–structure interaction. c, Prediction of ejected chamber volume (ΔRVV) and RVP during systole under physiological boundary conditions for both longitudinal and circumferential actuators. d, Prediction of ΔRVV and RVP during systole under physiological boundary conditions for circumferential actuators (no longitudinal actuators). e, Prediction of ΔRVV for various disease cases by selectively deactivating actuators and increasing the hemodynamic after-load. Disease cases 1 and 2 represent varying degrees of RV systolic failure, whereas disease case 3 recapitulates a pressure overload condition. f, Physiological hemodynamics are replicated in the RRV cardiovascular simulator, which produces pressure and flow waveforms that accurately…

• 

  Fig. 3. Comparison of in vivo and in vitro RRV hemodynamics. a, The RRV was constructed using a freshly explanted porcine heart. Hemodynamic data were collected from juvenile pigs in vivo and used to validate the accuracy of the in vitro model. b, The surgical and data acquisition setup to record invasive hemodynamic measurements in pig models. c, A mock circulatory flow loop incorporating hydraulic and mechanical components to simulate the pulmonary and systemic circulation on a laboratory bench. The RRV serves as the primary pump, driving flow through the loop. d, Right-sided pressure waveforms recorded in vivo. e, In vitro right-sided pressure waveforms were recreated using the RRV and the mock circulatory flow loop. f, In vivo flow waveforms show inflow and outflow. g, Representative flow waveforms recapitulated in vitro. Source data

• 

  Fig. 4. A comparison of in vivo and in vitro pressure waveforms. a, In vivo and in vitro pressure waveforms from the RA, RV and PA. b, The similarities between clinical indicators estimated from in vivo pig and in vitro right heart simulator hemodynamic data are highlighted using a web spider plot. Parameters include cardiac output (CO), mean pulmonary artery pressure (mPAP), pulmonary arterial compliance (PAC), stroke volume (SV), pulmonary transvalvular pressure gradient (PVPG) and pulmonary regurgitation (PR). Source data

• 

  Fig. 5. Imaging the motion and valve function of the RRV. a, The motion of the right ventricular free wall and intraventricular septum in the RRV is compared to in vivo porcine data, revealing similarities in the FAC (in vivo 56 ± 5% versus in vitro 50 ± 3%). b, The RRV model can be imaged using multiple modalities. An echocardiography image of the parasternal short-axis base view (left), which displays the inflow, outflow and trabeculated regions of the RV, and an MRI cine image (right), depicting the RV chamber’s circumferential contraction during diastole and systole, are used to demonstrate the RV’s anatomy and function. c, The functional PV and TV in the RRV are visualized using echocardiography (top) and an endoscopic camera (bottom).

• 

  Fig. 6. CFD models and MRI velocimetry. a, Computational fluid model to predict the interventricular fluid particle velocity. b, The magnitude and phase data of a 2D MRI flow measurement sequence are depicted. AP, anterior-posterior direction; FH, foot–head; RL, right–left. c, The cross-sectional data recorded during the study were carefully analyzed and corrected to accurately depict the distribution of flow across the RV. d, Reconstructed velocity vectors superimposed on the long-axis and short-axis cine images during both systole and diastole.

• 

  Fig. 7. The benchtop RRV can mimic various pathological conditions that affect the right side of the heart. a, RV volume overload is illustrated through TR. b, RV systolic failure is mimicked through a case example of RVMI. c, RV pressure overload is recreated by PAH. Source data

• 

  Fig. 8. A platform for testing right heart interventions. a, As an example, the benchtop RRV model is used to test devices for TR. In this scenario, a mechanical valve is substituted for the damaged valve. b, To replace the native valve, a surgical right atrial atriotomy is performed. c, TR before and after intervention, shown by flow measurements for three cycles and a regurgitant fraction calculated over 10 cycles, demonstrating that the ability to study both healthy and pathological tricuspid flow in the same model allows for more accurate and comprehensive comparisons of the impact of different interventions on the cardiovascular system. Data show average ± s.d. for 10 cycles (*P < 0.0001 (exact value: 1.18631 × 10−7), calculated using one-way ANOVA (Tukey test with a significance level of 0.05)). d, Surgical implantation of prosthetic annuloplasty rings (32 mm and 24 mm) at the tricuspid annulus through right atrial atriotomy. e, Endoscopic view showcasing leaflet coaptation before and after the annulus correction. f, Hemodynamic data illustrating improved tricuspid flow and RAPs after annulus correction. g, Integration of a McKibben-based soft robotic muscle into the RRV to…

• 

  Extended Data Fig. 1. Mechanical properties comparison. Comparison of the elastic mechanical modulus between formalin-fixed and decellularized myocardial muscle tissue samples, demonstrating the impact of the decellularization process on restoring tissue pliability and mechanical properties. Data shows mean ± stdev (n = 3 myocardial samples obtained from n = 1 porcine heart for each type). Source data

• 

  Extended Data Fig. 2. Computationally recreated pathology. Disease cases 1 and 2 represent systolic right ventricular failure and were computationally recreated by deactivating the actuators colored in red. Disease case 3 was created by increasing the time-dependent pressure boundary condition to 40 mmHg during systole.

• 

  Extended Data Fig. 3. Experimental fluid model. The setup of the mock circulatory flow loop.

• 

  Extended Data Fig. 4. Hemodynamic parameter tunability. Tunability of the pulmonary artery pressure (PAP) waveform in the simulator can be achieved by adjusting circuit parameters to simulate in vivo overdamping. This results in a PA pressure waveform that more closely matches the overdamped signal observed in vivo. Source data