PSI - Issue 49

S. McLennan et al. / Procedia Structural Integrity 49 (2023) 51–58 Author name / Structural Integrity Procedia 00 (2023) 000–000

53 3

2 Methods 2.1 Patient Data and Image Segmentation

From a database of 84 patients that presented with moderate to severe calcification presence, 12 were randomly selected for numerical simulation (9 male, 3 female; mean age, 75 years; range, 64-88 years). All 12 patients underwent EVAR between 2011 and 2018 at the Centre Hospitalier de l’Universit´e de Montr´eal (CHUM), Canada. All patients signed an informed consent form for being registered anonymously in the database. The study protocol was approved by the CHUM ethics committee. All EVAR procedures were on a stationary C-Arm (Artis zeego, Siemens Healthcare GmbH, Forchheim, Germany). Segmentation from CT scans was performed using a semi-automated 3D segmentation tool (ITK-SNAP 3.8) Yushkevich et al. (2006). The geometries were segmented for each patient in accordance with the methodology previously presented by our research group Mohammadi et al. (2018): abdomen, bone, intraluminal thrombus, and the aorta-iliac structure. An overview of the model components is provided in Figure 1. Aortic calcification for each patient was segmented using a 500 Hounsfield Units (HU) lower threshold. The threshold was applied to the CT scans prior to segmentation in order to adequately remove the intravenous contrast agent from the images, based on the recommendations of Buijs et al. (2018). To determine if our chosen 500 HU threshold captured an adequate volume of the aortic calcification, a sensitivity analysis was performed by using a lower threshold value of 400 HU on one of the patients. The calcification volume found using this value was then compared to the volume found when using 500 HU for the same patient. The 400 HU threshold increased calcification volume by only 4.26%. From this the 500 HU threshold was determined to be suitable. In the literature, de Weert et al. (2009) and Komen et al. (2011) also used a 500 HU threshold to respectively isolate carotid and iliac calcifications in C+ CT scans. 2.2 Finite Element Analysis FEA was performed using the LS-DYNA SMP R11.0 explicit finite-element solver (Livermore Software Technology Corporation, Livermore, CA, USA) on a 64-bit workstation with Intel(R) Xeon(R) CPU E5-2630 v3 using two 2.40 GHz processors. A detailed explanation of material property assignments, meshing methods, and general simulation procedures is toprovide in our previous work Mohammadi et al. (2018) and McLennan (2021). Based on the work of Maier et al. (2010), calcification was characterized a linear elastic material with a density of 0.12 g/mm 3 , Young’s Modulus of 50 MPa, and a Poisson’s ratio of 0.4. In order to include the calcification in the aortic wall, boundaries of the calcified tissue were identified and the areas of aortic wall within these boundaries were removed, allowing insertion of the calcification part. The calcification part was meshed using three-node triangular shell elements with a mean characteristic edge length of 1.5 mm. In addition, a tied contact was defined between the calcifications and wall using a node-merging operation. A mesh sensitivity analysis was performed to assess the adequacy of the 1.5 mm mesh edge length across the aorto-iliac structure. To do this we reduced the edge length size by half (0.75 mm) on all components of the aorto-iliac structure (aorta, intraluminal thrombus, and calcification) for one of the patients and re-ran the simulation. The PWS of the finer mesh simulation only differed by 1.78% to the original simulation. This difference was determined to be suitably small for justifying the 1.5 mm mesh edge on all patients. To assess the feasibility of EVAR for each patient using the numerical simulation model, we simulated the PWS of the non-calcified aortic tissue and compared it to the aortic wall strength as calculated using the empirical equation presented by Geest et al. (2006): ∗ � ��� � ����√ � ����� � ��������� � ����� � ������� � ������ (1) where σ ∗ is the aortic wall strength in units of kPa, ILT is the ILT thickness in units of cm, Nord is a dimensionless parameter for local normalized diameter, Hist is a dimensionless binary variable (1/2 for positive family history, -1/2 for negative family history), and Sex is a dimensionless binary variable (1/2 for males, -1/2 for females). If the simulated PWS exceeded the wall strength, we could make an assumption that EVAR was not feasible due to potential