14:30   Vascular II Chair: Mirunalini Thirugnanasambandam Ultrasound-Based Computational Fluid Dynamics Modelling of Healthy and Stenosed Carotid Arteries: Enhancing Understanding at the Site of Plaque Accumulation Lotte Piek, Joerik de Ruijter, Marc van Sambeek, Richard Lopata Abstract: Carotid artery disease, characterized by plaque buildup, poses significant risks such as strokes and transient ischemic attacks. Current interventions often rely solely on diameter measurements, which can lead to over- or undertreatment. To address this, patient-specific computational models are essential for more precise clinical decisions. This study aims to enhance the understanding of healthy and diseased carotid artery geometries using computational fluid dynamics (CFD), with the goal of supporting personalized treatment strategies. Ultrasound (US) scans of volunteers were segmented using a Star-Kalman algorithm (de Ruijter et al., 2020) , while patient scans were processed with a convolutional neural network (de Ruijter et al., 2021).The contours were then smoothened, and inlet and outlet extensions were applied. Inlet conditions were defined using clinical profiles, and downstream vasculature was modeled using Windkessel models. Blood was characterized by the Carreau viscosity model. Hemodynamic parameters such as time-averaged wall shear stress (TAWSS), oscillatory shear index (OSI), and relative residence time (RRT) were computed. These metrics were combined into a “RiskMap” to visually compare healthy and stenosed arteries in a single image. The RiskMap revealed that, in healthy arteries, the highest risk areas are located at the carotid bulb. However, in stenosed arteries, a greater and more widespread area of risk is evident, with higher maximum values. This indicates increased vulnerability in diseased geometries. Additionally, helicity was analyzed by examining time-averaged local normalized helicity (TALNH) and helicity parameters H1 to H4. TALNH showed lower magnitudes in healthy arteries compared to stenosed ones, indicating less rotational flow disturbance. Helicity parameter H1 was more negative in stenosed arteries, reflecting more chaotic flow. The higher H2 values in stenosed cases suggested greater variation in flow patterns. Parameter H3 was negative in diseased geometries, indicating a higher prevalence of negative (left-handed) helicity. Lastly, parameter H4 was higher in stenosed arteries, showing an increased magnitude of helicity overall. In summary, this study demonstrates that using CFD and helicity analysis in combination with ultrasound-based segmentation can improve the characterization of carotid artery disease. The findings suggest that combining these advanced hemodynamic metrics with RiskMaps may lead to better identification of high-risk areas. Wall Thickness in AAAs using IVUS In Vivo Floor Fasen, Marc van Sambeek, Richard Lopata Abstract: An abdominal aortic aneurysm (AAA) is a localized widening of the abdominal aorta. To prevent rupture, surgical intervention is needed. Currently, surgery is performed based on the diameter of the aorta. However, a better patient-specific marker is needed. The mean thickness of the wall is an indicator for rupture risk [1], which varies significantly within and between patients [2]. So far, regional thickness has only been measured on AAA specimens ex vivo [2]. This is the first study to map locally varying wall thickness of AAAs in vivo using intravascular ultrasound (IVUS). Pre-operative CTA scans were obtained from three AAA patients. During surgery, an automated IVUS pullback (1 mm/s) was performed. CTA and IVUS data were registered manually, from which the IVUS catheter path was retrieved. The wall in the IVUS images was automatically segmented, yielding wall thickness. After that, pulse cycles were retrieved, leading to a relative pulse cycle position for every IVUS frame. For every pulse cycle phase, a 3D geometry with locally varying wall thickness was obtained. In this study we showed locally varying wall thickness in AAAs, measured in vivo, which provides insight into the anatomy of the AAA wall. The results show a thinner wall in the dilated part (t1=2.0 ± 0.8 mm, t2=1.9 ± 0.7 mm, t3=1.7 ± 1.2 mm) compared to the non-dilated part (t1=2.5 ± 0.7 mm, t2=2.3 ± 0.6 mm, t3=2.1 ± 0.61 mm) within patients. Furthermore, in the dilated part, the posterior region (t1=1.6 ± 0.7 mm, t2=1.7 ± 0.8 mm, t3=1.7 ± 0.7 mm) is thinner than the left (t1=2.2 ± 0.7 mm, t2=1.9 ± 0.6 mm, t3=1.8 ± 0.6 mm), anterior (t1=2.0 ± 0.7 mm, t2=2.0 ± 0.7 mm, t3=2.2 ± 1.3 mm) and right (t1=2.0 ± 0.7 mm, t2=2.0 ± 0.7 mm, t3=2.2 ± 1.3 mm) region. These results agree with results previously found ex vivo [2]. Moreover, we successfully extracted 3D+t datasets from the pullback data. The proposed method can be used to obtain a patient-specific rupture risk marker in a minimally invasive manner. In future studies, mechanical properties of wall and thrombus will be investigated. In-vivo Analysis of the Circumferential Change of Z-Stent Graft Proximal Sealing Rings in Abdominal Aortic Aneurysms Patients. Elke Hestermann, Marleen Krommendijk, Hadi Mirgolbabaee, Erik Groot Jebbink, Michel Reijnen, Robert Geelkerken Abstract: An endovascular aortic repair (EVAR) procedure entails a stent graft being inserted into the aorta through an opening in the groin area to replace the wall of the aorta with the synthetic graft. Endoleaks are the most common complications of this procedure for abdominal aortic aneurysm (AAA) patients where the stent does not seal properly and blood still flows into the aneurysm sac which occurs up to 30% in patients 1. There are five types of endoleaks where type 1 endoleaks being the most dangerous and involves inadequate sealing at the proximal and distal areas 2. Therefore understanding the behaviour of the proximal sealing ring is important to ensure stability in the long-term and patient safety. The aim of this study is to investigate the dynamic behaviour of stent graft sealing rings to form a prospective analysis of the circumferential changes in both the proximal bare sealing ring and the first covered sealing ring after an extended post-deployment period. Thirteen ECG-gated scans of AAA patients will be analysed using a verified in-house Python algorithm, focusing on the proximal bare ring and first covered ring of the TREO™ (Terumo Aortic, Inchinnan, UK) stent graft. The 13 ECG-gated scans will span periods of 36 months (n = 4), 48 months (n = 7), and 60 months (n = 2). The nitinol wired rings of the TREO™ stent graft are in a symmetrical z-formation, and circumference measurements will be taken at the top and bottom nodes of each ring, as well as at each strut edge midpoint. The stent graft's circumferential difference will then be used to evaluate the expansion of the sealing rings throughout the cardiac cycle. By determining the minimum and maximum expansion of the sealing rings, the corresponding diametric distances will be calculated, enabling the final assessment of the pulsatility of the proximal sealing ring. Preliminary results indicates the sealing rings‘ stability increases over time as patients at 60 months showed little to none circumferential changed in comparison to the other. However, further investigation will provide insight into the dynamic behaviour of the sealing rings and the implications of oversizing on the aortic neck. Thrombus analysis for prediction of abdominal aortic aneurysm shrinkage after endovascular repair Rianne van Rijswijk, Dieuwertje Alblas, Jelmer Wolterink, Erik Groot Jebbink, Michel Reijnen Abstract: An abdominal aortic aneurysm (AAA) is a pathological widening of the abdominal aorta, which is often treated endovascularly (EVAR)1. Recent findings show that patients with shrinkage of the AAA one-year after EVAR have significantly better long-term outcomes than patients with a growing or stable AAA2,3. Knowledge of predictors of AAA shrinkage could therefore give insight into these preferred long-term EVAR outcomes. Studies on clinically used parameters have not yet identified strong predictors of AAA shrinkage, but small studies have hinted at involvement of intraluminal thrombus (ILT). Therefore, the current goal is to perform an in-depth preoperative ILT analysis to identify predictors of AAA shrinkage. In this two-centre retrospective cohort study, the lumen and ILT of the AAA were automatically segmented in the preoperative CTA-scan using an nnU-Net4. The lumen centerline was computed and multiplanar reconstructions of the lumen and ILT were created to enable measurements perpendicular to this centerline. A wide range of ILT-related parameters were automatically acquired from these multiplanar reconstructions, including the volume sizes and radiodensities (Hounsfield Units) of the lumen and ILT, the ratios of these values between the lumen and ILT, the ILT thickness, and the ILT circumference. Logistic regression was performed to find significant relations between ILT parameters and AAA remodeling. A total of 134 patients with a shrinking sac (≥5mm) one-year after EVAR were included, and 162 patients with a stable sac (<5mm). A multivariate logistic regression model indicated that a higher variation in ILT circumference, higher radiodensity of the ILT compared to the lumen, implantation of an Anaconda device, and younger age increased the likelihood of developing AAA shrinkage. Survival analysis confirmed that the long-term mortality rate was higher in patients with a stable sac compared to a shrinking sac. This study shows that automatic preoperative ILT analysis can aid in identifying patients likely to develop AAA shrinkage. In future work, the prediction could be improved by adding specific neck ILT measurements, calcification and other hostile neck parameters to the ILT analysis5. Overall, reliable prediction of AAA shrinkage can be used to improve EVAR treatment choice and enable patient-specific stratification of EVAR follow-up. Simulating Skin Sympathetic Nerve Activity: A Mathematical Model Runwei Lin, Frank Halfwerk, Dirk Donker, Gozewijn Dirk Laverman, Ying Wang Abstract: Altered neuro vegetative regulation may importantly impact on cardiovascular diseases, and can lead to impaired cardiovascular homeostasis. This has long been recognized, particularly in conditions like hypertension. Understanding such neurovegetative dysregulation might have important implications for the management for different cardiovascular diseases. Recently, skin sympathetic nerve activity has been measured noninvasively, and suggested as a very accessible parameter to quantify autonomic tone and its impact on a patient’s cardiovascular system. Moreover, its average value (aSKNA) has been proposed as a novel marker in clinical settings, including the screening of some cardiovascular diseases. In this study, we designed a model featuring the most relevant cardiovascular physiology to simulate heart rate changes and skin sympathetic nerve activity. To validate the model, we collected data while healthy subjects performing Valsalva manoeuvre (VM) to assess a patient's neurovegetaive regulation. The test can significantly decrease venous return and arterial blood pressure and in turn activate the sympathetic nervous system. Although the underlying mechanism is complex and in part potentially nonlinear, VM is modelled in a dedicated way to focus on the baroreflex and assuming a linear relationship between sympathetic tone and aSKNA. This approach aims to reduce model complexity while retaining the essential physiological mechanisms. Ethics approval was obtained at the Natural & Engineering Sciences Ethics committee of the University of Twente. A total of 41 healthy volunteers were enrolled in the experiment, where each subject participated in two measurements, performing two VMs in each. A burst analysis was utilized firstly to identify if the manoeuvre was performed correctly. Finally, a total of 37 measurements were analyzed. The average root mean square error between the predicted and measured aSKNA is 0.46μV and the Pearson correlation coefficient between them is 0.85. Figure 1. shows an example of the comparison between the predicted and observed aSKNA. In conclusion, the model is expected to improve the understanding of autonomic cardiac regulation during VM, as evidenced by the promising agreement between the simulation and measured aSKNA. It has the potential for further development in estimating autonomic cardiac tone in patients with autonomic dysregulation during Valsalva manoeuvre. This provides insights into the mechanism of relevant disease and presents opportunities for the detection of autonomic dysfunction. Towards advanced ultrasound simulations of virtual patients based on in silico microstructural tissue phantoms Daniek van Aarle, Jasper Korte, Richard Lopata, Hans-Martin Schwab Abstract: Medical ultrasound is a non-invasive image modality frequently used in the clinic offering real-time high-resolution imaging of soft tissues. A simulation method that accommodates tissue-specific scattering would significantly improve realism of in silico phantoms, generating urgently needed training data with ground truth information (on anatomy, motion, function) available. As the ultrasound wave propagates through tissue, it is crucial to include tissue microstructure into simulations, i.e., the spatial organization of fibers and cells on a micrometer scale, to achieve a realistic and accurate ultrasound image. In this research, we define an approach for the development of in silico tissue phantoms based on tissue microstructure. We demonstrate the approach on three different tissue types: carotid, muscle and skin. These are ultimately combined in a single simulation of a multilayered vascular tissue phantom, resembling the neck region (carotid). Ex vivo porcine tissue samples were processed for histological analysis using hematoxylin-picrosirius red staining. Conventional k-means segmentation was used to segment cells and collagen fibers. This segmented microstructure was low-pass filtered and down-sampled to a coarse simulation grid for simulation in k-Wave. Relative local acoustic properties (density and speed-of-sound) were estimated per tissue type by means of a forward simulation, comparing to ex vivo acquired ultrasound data of the same tissue. Quantitative evaluation was carried out on independent test samples in terms of Jensen-Shannon Divergence (JSD). Data were acquired with a Verasonics Vantage-256 system with L11-5v and L22-14v linear array transducers, using plane wave acquisition (41 angles, -20 to 20 degrees). Images were reconstructed using coherent compounding. A high resemblance is found for the ex vivo and in silico ultrasound images. Similar speckle patterns are observed, which is reflected in the low JSD values. The JSD (median with interquartile range) was found to be 0.05 (0.04-0.11) for the carotid, 0.03 (0.01-0.07) for the muscle, and 0.01 (0.006-0.02) for the skin. In future research, the method will be extended to generate parametric tissue phantoms to accommodate for the simulation of virtual patient ultrasound with a high level of realism.

end %-->