Intracranial aneurysms (IA) occur with a prevalence of 3-8% in the general population. The rupture of an aneurysm leads to subarachnoid hemorrhage, which accounts for 10% of all strokes and is fatal in 50% of cases. Numerous studies demonstrated that cerebral aneurysm progression and rupture can be affected by blood flow dynamics. Flow metrics such as wall shear stress, pressure and flow residence time were linked to local vessel wall remodeling and intra-aneurysmal thrombus deposition. Moreover, a detailed understanding of the complex, patient-specific flow dynamics is critical to identify clinically relevant flow descriptors for the planning of interventional procedures such as indirect aneurysm occlusion or implanting a flow diverter stent.Recently, 4D Flow MRI has been introduced for the in-vivo assessment of cardiovascular flow dynamics. This technique is capable of measuring time-resolved velocity fields in three dimensions. While there are reports on acquiring 4D Flow data in cerebral aneurysms, the small size of the cerebral arteries and limitations of 4D Flow MRI in both spatial and temporal resolution as well as velocity dynamic range affect the accuracy of these measurements. Inadequate resolution of the velocity fields can be detrimental for the estimation of flow-derived metrics such as pressure and shear stress. Moreover, to date, no rigorous error analysis of the 4D Flow MRI measurements was reported. Alternatively, the flow fields in cerebral aneurysms can be obtained from patient-specific models, either using Computational Fluid Dynamics (CFD) simulations or Particle Image Velocimetry (PIV) measurements. While providing much superior resolution in time and space, both CFD and PIV methods depend on modeling assumption and simplifications, which also affect the reliability of the resulting flow metrics. To overcome these limitations, we propose to develop a novel methodology that will allow for the spatiotemporal enhancement of 4D Flow MRI data to enable reliable quantification of clinically relevant flow metrics. We propose a multi-modality and multi-fidelity data-fusion approach to integrate 4D Flow MRI imaging with CFD and PIV modeling in order to reconstruct in-vivo 4D Flow MRI velocity measurements with spatiotemporal resolution and velocity dynamic range sufficient to compute hemodynamic metrics with high fidelity. In addition, we will assess the accuracy, uncertainty robustness and fidelity of the proposed methodology by comparing reconstructed velocity to highly controlled and accurate benchmark in-vitro phantom MR, CFD and PIV data which will serve as the reference standard. Ultimately, this research will deliver the foundations for a clinical tool for assessing the hemodynamics in cerebral aneurysms. The proposed framework will be a crucial first step in the future development of a deep learning approach for predicting aneurysm progression and rupture.The proposed study will include 1) Error analysis of 4D Flow MRI velocimetry in comparison to CFD and PIV in acquired healthy volunteer and patient 4D flow MRI data sets; 2) Data fusion of 4D Flow MRI, CFD and PIV modalities and 4D Flow MRI data enhancement of these data sets. The successful completion of the project will enable the development of a clinical tool for predicting cerebral aneurysm rupture on a patient basis and thereby for risk stratification of cerebral aneurysm patients.
|Effective start/end date||6/15/18 → 5/31/21|
- Purdue University (11000730-011//5R21NS106696-02)
- National Institute of Neurological Disorders and Stroke (11000730-011//5R21NS106696-02)