Ascending thoracic aortic aneurysms (ATAA) are associated with an increased risk of life-threatening dissection and rupture. Wall shear stress (WSS) is an important haemodynamic marker for aneurysm progression, but its computation using computational fluid dynamics (CFD) is computationally expensive, motivating the development of machine-learning (ML) surrogate models. Existing PointNet-based WSS surrogates predict WSS from vessel geometry alone and have not evaluated whether incorporating inlet velocity information improves prediction accuracy.

This study investigated the value of inlet velocity information through a controlled ablation study comparing two PointNet-based models: Model A, using only three-dimensional wall coordinates, and Model B, augmenting these coordinates with six scalar descriptors of the inlet velocity profile. Both models were evaluated using leave-one-geometry-out cross-validation across eight synthetic ATAA geometries generated from CFD simulations with varying inlet velocity profiles.

Model A consistently outperformed Model B across all evaluation metrics and folds, achieving a mean normalised mean absolute error (NMAE) of 7.63% compared with 13.44% for Model B, and a mean Pearson correlation coefficient of 0.49 compared with 0.19 (Wilcoxon signed-rank test, p = 0.0078). Model B additionally produced spatially inverted predictions in two out of eight folds, while both models systematically underestimated the highest WSS values.

The results demonstrate that, within the evaluated dataset and architecture, the tested six-scalar representation of inlet velocity did not provide additional predictive value and instead reduced WSS prediction accuracy. This finding does not imply that inlet velocity is unimportant for the underlying haemodynamics; rather, it suggests that compressed global velocity descriptors may be insufficient for improving geometry-based WSS surrogates. Future work should investigate spatially resolved velocity representations, larger and more diverse training datasets, and validation using patient-specific CFD data. ... Read more

Ischaemic stroke remains a leading cause of death and disability worldwide, and mechanical thrombectomy using stent retrievers is the current standard of care for its treatment. Bringing new thrombectomy devices to market, however, relies on costly and time-consuming preclinical and clinical trials that struggle to capture the full variability of patient anatomy and thrombus composition. In-silico clinical trials (ISCTs) offer a way to circumvent these constraints, provided the underlying device models are computationally efficient and demonstrably credible. This thesis, conducted within the T-SIM project, addresses this need by developing and validating a finite element model of the Trevo XP ProVue stent retriever, following the ASME V&V 10/40 framework.A high-fidelity solid-element (C3D8I) model of the stent geometry was constructed from microscopy measurements and converted into a computationally efficient beam-element (B31) representation. An initial comparison revealed discrepancies of up to 33% in stiffness properties between the two representations, traced to an under-stiffened connector–cell junction in the beam model; introducing an additional beam element at this location reduced the discrepancy to within 1–4%, while reducing the element and node count by more than an order of magnitude. The complete model, including end features and platinum–tungsten braided wires, was then validated against in-vitro flat-plate compression and three-point bending tests. The initial model (Version 1) showed poor agreement with experimental data (NRMSE of 36.0% and 27.0%, respectively); a sensitivity analysis identified strut width and depth as the dominant contributing parameters, motivating refined measurements and a revised model (Version 2), which improved agreement to 16.8% and 22.1% NRMSE. Neither version met the 10% acceptance criterion required under the model's assigned Risk Level 5 classification. Finally, four Context-of-Use simulations comparing the stent with and without the braided wires showed that the wires have negligible influence on radial mechanical behaviour but measurably restrict axial elongation during retraction.These results show that a computationally efficient beam-element model can reproduce the structural response of a high-fidelity solid model with good accuracy, but that full experimental validation against the required 10% threshold has not yet been achieved, leaving the central research question only partially answered. The findings nonetheless provide a reusable model development pipeline and a provisional simplification rule—omitting braided radiopacity wires for radial-loading applications—that can support the broader device library envisioned within the T-SIM project ... Read more

Vascular phantoms are artificial vessel models that provide controlled and repeatable environments for medical research, imaging validation, device testing, surgical planning, and training. To improve their physiological relevance, these phantoms should not only reproduce vascular geometry, but also show mechanical behaviour that is representative of blood vessels. In particular, small-scale compliant vascular phantoms should be able to deform under pressure-driven flow. However, fabricating small open channels from soft materials while maintaining geometrical accuracy, tunable stiffness, and measurable compliance remains challenging.

This thesis investigates the fabrication and experimental characterization of small-scale flexible vascular phantoms created using masked stereolithography 3D printing. Straight cylindrical channels were printed using Elastomer-X resin with different lumen diameters and wall thicknesses to determine a printable geometry range. The geometrical accuracy of the printed channels was evaluated using microscopy, while the internal lumen geometry was further inspected using ink-filled channels. Tensile tests were performed to determine the Young's modulus of the printed material under different exposure and post-curing conditions. By varying the printer exposure time and UV post-curing duration, different stiffness configurations were obtained.

The pressure-flow behaviour of the printed channels was investigated using water as the working fluid. Baseline measurements were used to correct for pressure losses in the setup, after which the channel-only hydraulic resistance was calculated. The effective diameter was then estimated from the hydraulic resistance using the Hagen--Poiseuille relation. The results showed that a lumen diameter of 0.7~mm with a wall thickness of 0.25~mm could be printed as the smallest consistently open channel suitable for flow experiments. Tensile testing showed that the Young's modulus could be tuned between approximately 1.05 and 1.25~MPa by changing the fabrication conditions. Pressure-flow measurements showed a decrease in hydraulic resistance with increasing mean pressure for several channels, corresponding to an increase in effective diameter. The softer channels generally showed a larger relative effective diameter increase than the slightly stiffer channels.

These results demonstrate that masked stereolithography can be used to fabricate small-scale flexible vascular phantoms with tunable stiffness and measurable pressure-dependent behaviour. Although the effective diameter was determined indirectly and uncertainties remain due to local lumen narrowing, measurement noise, and baseline correction, the study provides an important experimental validation step towards compliant 3D printed vascular phantoms for pressure-flow applications. ... Read more

Physician-modified endografts (PMEG) are increasingly used for the endovascular treatment of complex abdominal aortic aneurysms, particularly in urgent clinical settings where custom-made devices are not available. Despite their growing use, current PMEG fenestration practices inherently show substantial variability in reinforcement design and execution, while experimental characterization remains fragmented, raising risks for patient safety and limiting engineering-driven
optimization of designs. This study presents the design and feasibility evaluation of standardized fenestration reinforcement concepts using a structured and fenestration centered testing framework. Two adhesive based reinforcement strategies, the RingPatch and the Patch, were developed using a voice-of-customer driven design approach and benchmarked against an established suture based reference, the KeyRing. Performance was evaluated using an ISO-informed feasibility testing workflow incorporating bridging stent graft deployment, junction seal testing, cyclic fatigue simulation, and pull-out force testing. The RingPatch demonstrated inadequate performance due to reinforcement detachment associated with insufficient adhesive curing and stiffness mismatch, whereas the Patch consistently exhibited superior junction integrity, sealing performance, and mechanical durability across all tests, outperforming the KeyRing reference design. These results demonstrate the ability of the proposed framework to identify dominant failure mechanisms and discriminate between competing reinforcement strategies at an early design stage. This study contributes a structured foundation for the development and future standardization of mechanically robust PMEG fenestration designs. ... Read more

Post-myocardial infarction (MI) growth and remodeling (G&R), commonly referred to as fibrosis, involves both geometric deformation and progressive stiffening of infarcted tissue due to collagen accumulation. While zero-dimensional (0D) cardiac G&R models have successfully reproduced organ-level adaptations post-infarction, they often neglect evolving tissue properties associated with collagen turnover. In this study, we address this limitation by incorporating a time-dependent stiffening law into a strain-driven 0D framework, extending the original model by Witzenburg et al. (2018). Collagen turnover (CT) was modeled using a phenomenological exponential function, calibrated against experimental hydroxyproline data. The model was validated against independent canine datasets and benchmarked against both the original reference and a baseline No CT simulation. While full time-course verification was not achieved - due to inconsistencies in baseline reported parameters - control and acute states were accurately reproduced. Critically, the CT-enhanced model reduced the mean standardized z-score (MSZ) by 57.8%, with the most substantial improvements seen in ventricular volume and diastolic pressure predictions. These results confirm the added value of explicitly modeling tissue-level remodeling and highlight the importance of accurate initialization to ensure long-term prediction fidelity in reduced-order frameworks. ... Read more

Objective Action tremors affect many Parkinson’s disease (PD) patients and can significantly impair daily functioning. These tremors often persist despite dopaminergic therapy, while stereotactic surgery, though effective, is invasive and not suitable for all. Alternative treatment options are therefore needed. The STIL Orthosis is a wearable device developed to reduce forearm tremor. Its effectiveness has been shown in essential tremor, but its impact on PD-related action tremors remains unknown. This pilot study evaluates its effect in a small PD sample.

Methods: Three PD patients with action tremor participated in a sham-controlled, single-blind, crossover pilot study. Tremor severity was assessed in three conditions: baseline, sham, and STIL Orthosis. The primary outcome was the TETRAS score. Secondary outcomes included MDS-UPDRS tremor scores, tremor power (accelerometry and sEMG), and patient satisfaction (PGI-I).

Results: Although effects did not reach statistical significance (for n=3), a consistent trend toward tremor reduction was observed across clinical and accelerometry-based measures when comparing the STIL Orthosis to baseline. TETRAS scores decreased from baseline \(13 \pm 4.0\) and sham \(11 \pm 5.2\) to orthosis \(8.0 \pm 2.1\). MDS-UPDRS tremor scores similarly declined (baseline: \(7.3 \pm 0.9\), sham: \(5.3 \pm 1.7\), orthosis: \(3.3 \pm 0.9\)). Two participants showed reduced tremor power (up to \(55\%\) compared to baseline). EMG tremor power of the ECR and FCR muscles decreased by \(15\%\) and \(58\%\), respectively. All three participants reported slight improvement with the orthosis (PGI-I).

Conclusion: This pilot study indicates that the STIL Orthosis influences action tremor in PD. Larger studies are needed to assess its clinical significance. ... Read more

Nasal airway obstruction (NAO) is one of the most common symptoms in the human respiratory system and causes a considerable financial burden to both individuals and society. Currently, NAO detection is troublesome to achieve, which can influence the accuracy and effectiveness of clinical surgery. Although several objective measurement techniques are currently available, they are found to be inconsistent with patients’ sensations. In recent years, computational fluid dynamics (CFD) has become a novel technique to objectively assess nasal airflow by simulating the nasal airflow of NAO patients. Existing literature has also shown the potential to correlate relevant CFD parameters with patients’ sensations. Nevertheless, there is still debate on the numerical setup for an accurate solution and the CFD parameter to correlate with the subjective measurement. Based on these existing issues, we mainly investigated the boundary configuration (with and without the external nose), the usage of turbulence/laminar models, the usage of steady-state/transient solvers, and briefly discussed the potential of using unilateral pressure drop ratio as a parameter for NAO detection. 

To begin with, including the external nose in the nasal airflow simulation is recommended in the nasal airflow simulations. The external nose configuration can affect the flow direction through the nostrils and downstream flow distributions. However, the static (and total) pressure drop only shows a 4% difference compared to the commonly-used plane-truncated boundary configuration. Furthermore, using the laminar model is sufficient for the nasal airflow simulations concerning the static pressure drop prediction. The laminar model shows a difference lower than 15% in static pressure drop compared to the experimental values on 3D-printed nasal airway models. We stress the caution of using the 𝑘 − 𝜔 model in the nasal airflow simulations because it tends to overpredict the turbulent viscosity ratio near the inlet unphysically. Moreover, steady-state simulations can also reasonably predict nasal airflow. We observed unsteady effects when comparing the steady-state simulation with the transient simulation with a constant flow rate and the transient simulation with a sinusoidal flow-rate-versus-time profile representing the real-life breathing cycle. Nevertheless, the steady-state simulation achieves an accurate prediction in static pressure drop, with a difference lower than 6% compared to the tested transient simulations. The steady-state simulation can also perfectly match transient simulations in the velocity profile of the recirculation zones and require a much lower computational cost. Last but not least, we also tested the possibility of using the unilateral static pressure drop ratio for NAO detection using CFD. However, we note that future studies should make corrections to account for the nasal cycle effect for NAO detection. 

Overall, we conclude that including the external nose and using the laminar simulations with the steady-state solver can give an acceptable prediction for the nasal airflow, especially concerning the static pressure drop prediction. We also state that applying CFD in the nasal airflow shows the potential for NAO detection, although future studies may consider making some corrections to include the nasal cycle effect. ... Read more