Patient-specific Planning of Minimally Invasive Brain Interventions Based on Vascular-hemodynamic Mapping

Patient-specific Planning of Minimally Invasive Brain Interventions Based on Vascular-hemodynamic Mapping

Endovascular procedures for treating brain aneurysms and arteriovenous malformations (AVMs) are becoming increasingly popular due to their less invasive nature and lower risk of complications. However, they still face challenges such as difficult catheter navigation and incomplete embolization.

This study aims to improve the efficiency and safety of endovascular procedures by developing new preoperative planning methodologies. These methodologies involve mapping the cerebral vasculature and creating computational fluid dynamics (CFD) and artificial intelligence (AI) models to simulate blood flow. By using these models, interventional radiologists can better plan catheter navigation and predict embolization outcomes. This could lead to faster, more accurate procedures with reduced radiation exposure for patients.

The number of treatments of blood vessels using a catheter inserted into a blood vessel, also known as endovascular treatment, has been increasing over the past decades. There are several advantages associated with this technique, such as that it is less invasive and carries a lower risk of complications. This technique is expected to become the preferred treatment for intracranial aneurysms and arteriovenous malformations (AVMs).

Despite the advantages of endovascular procedures, there are still technical challenges that can affect their effectiveness and safety. In clinical practice, a guidewire (approximately 1.5 to 2.0 m) is directed under fluoroscopic imaging to the specific brain area. Accurately navigating the tip of a passive guidewire by manipulating the proximal end requires many years of experience and can be a cumbersome task even for experienced interventional radiologists. Moreover, navigating the catheter to the target brain area can already take up a large part of the procedure time. A second challenge is the risk of incomplete embolization or recanalization. The risk of incomplete embolization is particularly present in AVMs, as they have a complex structure. The performance of an embolization can cause the hemodynamic in that particular brain region to change, resulting in other preferential flow paths, which can later lead to another embolization procedure. This creates a need for repeated angiographic evaluation, which also leads to longer and more expensive procedures and higher exposure to radiation for the patient. The above challenges show that there is a need for a method that allows for faster and more efficient neurological interventions, with the patient receiving a lower dose of radiation and a higher probability of complete and lasting embolization.

The goal of the study is to develop a new methodology for preoperative planning of catheter navigation and embolization, based on mapping of the cerebral vasculature and modelling of blood flow through the associated arterial vasculature. In cardiology, there are already techniques for guided navigation of the guidewire or other tool without the need for fluoroscopic guidance. In these techniques, the patient-specific vasculature is mapped, with a detailed 3D representation of the vascular network displayed together with tracking of the position of the catheter relative to this map in a correct 3D reference frame. However, these techniques have not yet been extended to neurovascular interventions, since the cerebral vasculature is an intrinsically complex arterial vasculature.

For the current study, there are two main objectives. The first primary objective of this research is to map the cerebral vasculature by quantifying and analysing 3D angiographic data with the aim of assessing the accessibility of the vasculature for catheter navigation and supporting catheter tracking.

The second main objective is to develop and validate computational fluid dynamics (CFD) and artificial intelligence (AI) models of cerebral blood flow, with the aim of supporting preoperative planning. These AI models serve to accelerate the CFD models. The clinical aspect of this study only involves the collection of retrospective and prospective patient data, so there is no additional clinical risk for the patient. Preferably, medical images (cerebral MRI images (resolution: 0.3 - 0.5 mm), 3D rotational angiography (3DRA, resolution: 0.05 - 0.28 mm), biplanar digital subtraction angiography (DSA, resolution: 0.15 - 0.62 mm)), together with a case report form (CRF, including therapy parameters and response). After data collection, patient-specific models of the (accessibility of the) vasculature and fluid distribution in the cerebral blood vessels will be developed on the side of UGent (research groups IbiTech BioMMedA; MEDISIP, Department of Electronics and Information Systems and research group IPI, Department of Telecommunications and Information Processing, both belonging to the Faculty of Engineering and Architecture at UGent). Medical images play a crucial role in the development and validation of these models. Once these models have been validated (including using in vitro models with 3D-printed phantoms), they may also be used in clinical applications in the longer term. Validation of these models in a clinical, interventional setting is not yet applicable for this study.

Inclusion Criteria:

• Subjects who are diagnosed with aneurysm or arteriovenous malformation
• Subjects who are treated (or will be treated) with endovascular embolization and/or arteriovenous coiling
• During and before/after the subject's intervention, medical images (MRI, 3D rotational angiography, DSA) were collected of the cerebral arterial vascular tree (which can be used to develop patient-specific models), including catheter location and use of contrast fluid to visualize the cerebral to make the arterial tree more visible.

Exclusion Criteria:

-Procedures during which no imaging was performed