How do we simulate EVAR before the intervention?
What is Aortic Digital Twin?
From a patient’s pre-operative CT scan, the aortic digital twin is created to faithfully reproduce both the shape and the bio mechanical behaviour of the real aorta. This advanced 3D model allows precise simulation of how the patient’s anatomy responds to mechanical forces, providing a reliable virtual counter part of the vessel.
To ensure accuracy, PrediSurge has conducted multiple experimental studies assessing the mechanical properties of the aorta. A representative study is described by Duprey et al. (2016, Acta Biomaterialia), which investigated the rupture and deformation behaviour of aortic samples under multi axial loading. These experiments provide critical reference data for modelling tissue strength and stretch responses before rupture.
How do we simulate EVAR implantation?
Step 1: Creation of Digital Twin of Aorta
The first step involves creating a patient-specific digital twin of the aorta. The aorta is segmented from a contrast-enhanced pre-operative CT scan, following a process similar to many other sizing software tools.
This aortic model is then meshed — a mathematical process that divides its surface into thousands of small elements. Each element is assigned the bio mechanical properties of the aortic wall.
As a result, the model transitions from a static image to a numerical entity capable of interacting with tools and deforming according to their mechanical properties.
This technology – unique to Predisurge - is the digital twin of the patient’s aorta.
Step 2: Digital Replica of Endograft
The second step consists of building a digital replica of the endograft. A detailed model reproducing the geometry of the stents and fabric is constructed and meshed.
The known stiffness and mechanical behaviour of both components are then applied to this digital replica of the endograft.
Step 3: Simulating the implantation of Endograft
Finally, the third step simulates the implantation of the endograft within the patient-specific aortic digital twin. The ERI user (either clinical specialist or the vascular surgeon) selects the endograft type, the introduction side, and defines the proximal landing zone relative to a reference artery (for example, 2 mm below the lower edge of the right renal artery).
The digital endograft is virtually crimped, introduced into the aortic model following these parameters, and then released. It interacts with the aortic digital twin until a mechanical equilibrium is reached — a state that corresponds to what is typically observed on post-operative CT scans.
This EVAR simulation can be used similarly to a post-operative CT scan. In particular, it allows for a detailed analysis of the endograft apposition. This is precisely what is automatically performed by Endoleak Risk Index.
How does the simulation take into account the way the operator will deploy the stent-graft?
Proximal landing zone of the stent graft is actually an input defined by the clinical specialist or the physician as a distance to a reference artery, usually the lower edge of the lowest renal artery. If the operator is not sure about thelanding zone reachable, multiple proximal landing zones scenarios can be simulated.
Validation of EVAR simulation
To validate the accuracy of EVAR simulations, retrospective studies were performed on patients who had previously undergone EVAR. For each case, the pre-operative CT scan was used to generate a patient-specific digital twin of the aorta. The procedure was then simulated virtually using the same Endurant endograft reference and identical landing zones as those employed during the actual intervention.
The predictive simulations were evaluated against post-operative CT scans. Specifically, the stent rings on the post-operative imaging were extracted and compared to the corresponding stent rings in the simulated digital twin.
Their position and their diameters were compared to evaluate the accuracy of the simulation, as described by Perrin et al. (Journal of Biomechanics, 2015) and Derycke et al. (Computer Methods and Programs in Biomedicine, 2024), providing a robust framework for assessing model fidelity.
The comparison demonstrated excellent agreement between simulated and real-world results. The position and expansion of the stent rings in the simulations closely matched those observed on patients' post-operative scans.
On average, the difference in stent ring radius between simulated and actual devices was less than 1 mm, highlighting the accuracy of the simulation.