Research

My research builds predictive, patient-specific models of cardiovascular disease and tissue-engineered devices, fusing mechanistic simulation with data-driven discovery.

Tissue-engineered vascular grafts (TEVGs)

At Stanford, I lead hybrid mechanistic + AI/ML workflows to optimize TEVG design by integrating large-animal experimental data with patient-specific FEA / CFD / FSI simulations. The goal is to predict graft remodeling outcomes and inform scaffold design before reaching the clinic.

Scientific machine learning for soft-tissue mechanics

I develop interpretable constitutive models for native and engineered vascular tissue using neural-network-based discovery (CANNs) and symbolic regression. By combining Sobol sensitivity analysis with these methods, learned models are simplified into closed-form strain energy functions that integrate directly into FEA workflows.

Multimodal imaging pipelines

I build computer-vision pipelines for intravascular ultrasound (IVUS): frame extraction, probe masking, contour segmentation and smoothing, 3D reconstruction, and STL/mesh generation. The output is subject-specific geometry that drives FEA and CFD simulations within SimVascular.

I also work with MRI and catheterization data, harmonizing them into pipelines that feed both mechanistic models and downstream ML.

Multiscale modeling of vascular disease

During my Ph.D., I developed multiscale predictive models of:

  • Aortic aneurysm progression — including sex-specific dilation in Marfan and Fbln4-SMKO mouse models
  • Patient-specific blood rheology — particularly in sickle-cell disease
  • Intracranial aneurysm oxygen transport — coupling flow, transport, and disease physiology

These were implemented in C++, Python, MATLAB, and Ansys, and deployed on Linux/SLURM HPC environments.

Open-source contributions

I contribute to the SimVascular ecosystem (C++ / Fortran), including the svFSI, svFSG, and svMultiPhysics solvers — adding modules for mesh generation, soft-tissue constitutive mechanics, and fluid–solid–growth multiphysics coupling.