• Atrial Fibrillation
• Blood Stasis
• Clinical Images Segmentation
• Fluid Structure Interaction simulations
• Left Atrial Appendage
• Left Atrium
• Risk of Strokes
• Shear Strain Rate
• Statistical Shape Analysis
• Thromboembolic Potential

Atrial fibrillation (AF) is one of the most common arrhythmias worldwide and it is estimated to affect almost 18 million people in Europe by 2060. AF is characterized by irregular, rapid activations of the atria and by anatomical remodeling, leading to blood stagnation, which increases the risk of clots formation and strokes. In AF patients, more than 90% of the strokes are caused by thrombus generated in the left atria appendage (LAA) works as a decompression chamber during left ventricular systole and during other periods when left atrial pressure is high. Its shape varies significantly among patients, and the current LAA classification system categorizes it into 4 morphologies: Chicken Wing, Windsock, Cactus, and Cauliflower. Recently, it has been demonstrated that there is correlation between LAA’s geometry and risk of thrombus formation.
Current therapies for strokes prevention are anticoagulant, percutaneous left atrial appendage occlusion and surgical left atrial appendage exclusion. However, each of them is characterized by downsides. Therefore, an improved patients stratification in term of thromboembolic risk potential could allow to choose the best therapy for each clinical case. For this purpose, in the presented study, CT-images of 46 AF patients were analysed and their 3D anatomical representation (including LA and LAA) was obtained through segmentation algorithms. The LAA was the focus of the study, thus it was cut off from the atrium. Statistical Shape Analysis was carried out to obtain the average shape (i.e. “template”). Furthermore, four different geometries were constructed deforming the template along the two first modesrepresenting the 30.3 % of the total variance. Fluid Structure Interaction (FSI) computational simulations were performed for the four cited representative LAA shapes.: these were initially meshed considering the inflation to solve the boundary layer flow and an external mesh to represent the left atrial appendage wall (LAAW). The inflation was composed by five hexahedron layers for the exterior part, while the inner part was meshed with tetrahedron elements.
Ansys CFX and Transient Structural were adopted for the simulations. The blood was treated as homogenous, Newtonian and incompressible fluid and a linear elastic modulus was used for the LAAW. In order to simulate the expansion and contraction of the LAA during the cardiac cycle, a thermal expansion strategy was adopted by setting an orthotropic thermal coefficient. As far as the CFX boundary conditions are concerned, an opening condition was set at the inlet, and a wall condition was considered for the wall. In the transient structural, three boundary conditions were chosen, a thermal condition over the whole domain, a fixed support on the inlet surface and a fluid solid interface condition was set on the inner wall.
The results were analyzed in terms of Sher Strain Rate (SSR), velocity streamline, wall displacement. SSR was assessed over the whole geometries in order to establish which one was the most critical. In fact, SSR values below 5 s^(-1) are related with increased risk of thrombus formation. For this reason, SSR average among the third cardiac cycle was computed and the percentage of nodes with SSR below 5 s^(-1) was calculated. The evaluation of the LAAW displacement showed which shape was characterized by the lowest movement.
Finally, a comparison between sinus rhythm and AF condition was carried out, simulating AF by imposing a rigid wall (i.e. standalone CFD). The comparison underlined a reduction in terms of SSR and velocity streamline for the AF case.
The presented work aimed at evaluating thromboembolic potential among different shapes through computational analysis. Additionally, it showed the principal differences between normal and pathological conditions in terms of hemodynamic parameters.