Numerical modeling of atrial dynamics and venous cannulation for extracorporeal life support
Time: Mon 2026-01-26 13.00
Location: D3, Lindstedtvägen 5
Language: English
Subject area: Engineering Mechanics
Doctoral student: Hanna Hörwing , Strömningsmekanik
Opponent: Prof. Gabriele Dubini, Politecnico di Milano
Supervisor: Prof. Lisa Prahl Wittberg, Strömningsmekanik; Docent Lars Mikael Broman, ; Dr. Louis Parker,
QC 251216
Abstract
Venovenous extracorporeal membrane oxygenation (ECMO) is a life saving therapy for critically ill patients with refractory respiratory failure. To support gas exchange, blood is drained from the patient via a cannula and circulated outside the body through a membrane lung, after which the blood is returned to the patient. The exposure of blood to non-physiological conditions triggers blood damage and thrombus formation in the circuit. While cannula placement induces different blood flow dynamics, and thus distinct complications, the decision of cannula configuration often lands on the preferences of the medical center.
From medical imaging, a patient-averaged model was derived and cannulae were inserted in femoro-femoral (FF) configuration. Previously studied configurations (femoro-jugular and jugulo-femoral) were re-simulated in the updated geometry. Through large eddy simulations (LES), a direct comparison of configurations could be made. The focus was specifically on clinically relevant metrics for oxygenation performance, thrombus formation and blood damage. FF generated more pronounced negative pressures in the inferior vana cava, associated with a risk of vessel collapse. Wall shear stresses, linked to thrombosis and plaque formation, exceeded recommended limits even at low ECMO flow rates. Furthering our understanding of induced flow dynamics from cannulation by such means may provide insights on optimal cannulation strategies, aiding clinicians in making informed decisions.
Exact modeling of vascular walls and blood is challenging due to the complexity and variability of the cardiovascular system, which has led to a common modeling simplification being treating walls and heart chambers as rigid. While this assumption facilitates computational modeling, it is important to assess its validity. A dynamic model of the right atrium (RA) was thus created using mesh morphing, mirroring known motion of the atrial wall. Through LES simulations, hemodynamic metrics were compared to a rigid counterpart model, establishing a sensitivity assessment of the rigid wall assumption for RA modeling. The rigid model underestimated fluid activity in the auricle, leading to an underestimation in wall shear stress and an overestimation of blood residence time and stagnation in this region. These results provide guidance on the validity of the rigid wall assumption, to increase our understanding of which sensitivities are important to consider for specific modeling applications.