Embedded Files
Abstract:
KD is a rare acquired heart disease in children characterized by coronary artery aneurysms with risks of thrombosis, myocardial infraction and sudden death. The growth of coronary artery aneurysms plays an important role in the anticoagulation therapy. Current clinical guidelines recommend CAA diameter ≥ 8 mm or Z-score >10 as the criterion for initiating systemic anticoagulation therapy. However the diameter based criterion remain unsatisfactory in the risk stratification.
We have collaborated with Dr. Burns, a leading expert in KD at University of California San Diego on modeling hemodynamics in children with KD and developed a non-invasive modeling tool to characterize abnormal coronary flow for individual patients. Our previous studies on modeling Kawasaki disease and adult coronary artery diseases equip us with necessary modeling techniques for this emerging problem. Image based flow simulations and automated parameter tuning will be employed.
Funding Sources: National Institute of Health, Private fundings.
Collaborators: Karthik Menon (Stanford), Alison Masrden (Stanford), Jane Burns (UCSD)
Automation of the patient-specific modeling building process using the machine learning and parameter estimation
We have successfully automated the whole patient-specific modeling process to enable a large size cohort study.
We have automated the modeling building process by combining a 3D segmentation method and the machine learning module, and achieved a speedup of roughly 5-10 times of the patient-specific modeling process, compared to the traditional approach.
This works has been intergrated in the Simvascular project: http://simvascular.github.io/docsModelGuide.html#modeling3DSeg
The machine learning framework is described in this paper: Maher et al. "Accelerating cardiovascular model building with convolutional neural networks", 2019, Medical & Biological Engineering & Computing (https://link.springer.com/article/10.1007/s11517-019-02029-3)
We have implemented the automatic boundary condition tuning approach for assigning physiologic information to the computation.
The framework refers to this paper: Tran et al. "Automated tuning for parameter identification and uncertainty quantification in multi-scale coronary simulations", 2019, Computers and Fluids (https://www.sciencedirect.com/science/article/abs/pii/S004579301630161X?via%3Dihub)
Using the automated modeling pipeline, our team members have generated new 13 computational KD patient models. We are planning to generate up to 40 patient models.
Computational evaluation of the fractional flow reserve (FFR) of coronary arteries in Kawasaki disease patients
We have evaluated the fractional flow reserve of the Kawasaki disease patients using the computational models. We examine the correlation between the FFR and the aneurysm geometry, and other hemodynamic quantities.
Report abuse