Abstract
Aortic dissection is a complex unsolved issue in medical and biomechanics fields. It affects the main artery of the human body, the vessel that carries oxygenrich blood from the heart to the rest of the body through branch arteries: the aorta. The dissection of the aorta occurs inside its wall, which is composed of three different layers of material, i.e., intima, media, and adventitia. The loadbearing structure of the aorta is the media, which is composed of 50 to 70 sublayers. The separation of the media layers and consecutive rupture of the intima causes the blood to flow into the newly formed cavity. The blood pressure will cause the propagation of the secondary volume, creating the aortic dissection. The new volume, in which the blood flows, is called a false lumen. According to the Stanford Classification System, aortic dissection initiates from the origin of an initial tear. A relative position to the ascending aorta will classify the disease as Type A aortic dissection or Type B aortic dissection.
Consequently, a profound study on aortic wall mechanics and fluidstructure interaction are at the basis of the mechanics of the disease. As for the hemodynamics of the system, it is worth mentioning the beneficial role of thrombus formation in such a disease. The hemodynamics condition in the false lumen often promotes the formation and growth of thrombi. Thrombus development in aortic dissection has been the focus of many medical studies. The thrombus was found to have a beneficial effect on a patient’s prognosis. The disease’s incidence is about 3 to 6 cases per 100000 population in its acute condition. However, recent studies mention numbers up to 15 cases per 100000 population due to better diagnostics and awareness of the disease among the medical community. Nonetheless, the disease’s management is challenging due to its rapid progression, given the high uncertainty in its diagnosis and treatment. At the moment, it is still unclear what is the mechanism that initiates aortic dissection, how and how fast it propagates, and which is the best treatment to cure it.
Aortic dissection is a medical problem that benefits from the engineering approaches, like many other scientific issues. The medical field is strongly affected by high uncertainty due to the extreme variation of the human body. Here, body proportions, the morphology of internal organs, lifestyles, and living environment represent some of the challenges that modelers face when reproducing the mechanics of the human body through complex computational models. The material and failure properties of aortic walls dominate the occurrence and progression of a tear in the aorta. Numerical values of the parameters are inherently difficult to determine and include a certain level of epistemic uncertainty, i.e., the tissue may be altered by genetics, loading conditions, or traumas. Hence, the location of material and geometrical imperfections can hardly be predicted.
Sensitivity analysis is used to investigate the variations in the output of a dynamical system caused by input values. Among various approaches, global sensitivity analysis with the variancebased method allows the analyst to focus on output variations. This approach also examines the whole input space of a model considering possible interactions among the input factors. On the contrary, a local approach will only investigate how the variation of single parameters will change the quantity of interest of the numerical model. A variancebased approach requires relevant statistical information, for instance, the probability density function of the desired model output. Given the vast amount of data that sensitivity analysis requires, the use of metamodels, such as polynomial chaos expansion, is considered. Metamodels allow the computation of more or less expensive numerical simulations and are convenient for quantitative sensitivity measures.
Sensitivity analysis is applied to reduce the uncertainty in aortic dissection detection using impedance cardiography, a noninvasive methodology primarily aimed at the medical community. Furthermore, given the lack of knowledge on the thrombus formation model, sensitivity analysis assists the process of defining the governing variables that could accelerate or inhibit its formation and growth. The use of sensitivity analysis helps to reduce the uncertainty in the modeling phase, aids the modelers in the model reduction process, and promotes the bond between engineers and physicians. While the firsts are more concerned with a mathematical and computational approach to the disease, developing new models and predictors demands a clearer understanding of the mechanics of the process and the applicability of newly developed tools.
Consequently, a profound study on aortic wall mechanics and fluidstructure interaction are at the basis of the mechanics of the disease. As for the hemodynamics of the system, it is worth mentioning the beneficial role of thrombus formation in such a disease. The hemodynamics condition in the false lumen often promotes the formation and growth of thrombi. Thrombus development in aortic dissection has been the focus of many medical studies. The thrombus was found to have a beneficial effect on a patient’s prognosis. The disease’s incidence is about 3 to 6 cases per 100000 population in its acute condition. However, recent studies mention numbers up to 15 cases per 100000 population due to better diagnostics and awareness of the disease among the medical community. Nonetheless, the disease’s management is challenging due to its rapid progression, given the high uncertainty in its diagnosis and treatment. At the moment, it is still unclear what is the mechanism that initiates aortic dissection, how and how fast it propagates, and which is the best treatment to cure it.
Aortic dissection is a medical problem that benefits from the engineering approaches, like many other scientific issues. The medical field is strongly affected by high uncertainty due to the extreme variation of the human body. Here, body proportions, the morphology of internal organs, lifestyles, and living environment represent some of the challenges that modelers face when reproducing the mechanics of the human body through complex computational models. The material and failure properties of aortic walls dominate the occurrence and progression of a tear in the aorta. Numerical values of the parameters are inherently difficult to determine and include a certain level of epistemic uncertainty, i.e., the tissue may be altered by genetics, loading conditions, or traumas. Hence, the location of material and geometrical imperfections can hardly be predicted.
Sensitivity analysis is used to investigate the variations in the output of a dynamical system caused by input values. Among various approaches, global sensitivity analysis with the variancebased method allows the analyst to focus on output variations. This approach also examines the whole input space of a model considering possible interactions among the input factors. On the contrary, a local approach will only investigate how the variation of single parameters will change the quantity of interest of the numerical model. A variancebased approach requires relevant statistical information, for instance, the probability density function of the desired model output. Given the vast amount of data that sensitivity analysis requires, the use of metamodels, such as polynomial chaos expansion, is considered. Metamodels allow the computation of more or less expensive numerical simulations and are convenient for quantitative sensitivity measures.
Sensitivity analysis is applied to reduce the uncertainty in aortic dissection detection using impedance cardiography, a noninvasive methodology primarily aimed at the medical community. Furthermore, given the lack of knowledge on the thrombus formation model, sensitivity analysis assists the process of defining the governing variables that could accelerate or inhibit its formation and growth. The use of sensitivity analysis helps to reduce the uncertainty in the modeling phase, aids the modelers in the model reduction process, and promotes the bond between engineers and physicians. While the firsts are more concerned with a mathematical and computational approach to the disease, developing new models and predictors demands a clearer understanding of the mechanics of the process and the applicability of newly developed tools.
Original language  English 

Qualification  Doctor of Technology 
Awarding Institution 

Supervisors/Advisors 

Publication status  Published  13 Apr 2022 