Transfer learning on PINNs for tracking hemodynamics

The source code and data for the article M. Daneker, S. Cai, Y. Qian, E. Myzelev, A. Khumbhat & L. Lu. Transfer Learning on Physics-Informed Neural Networks for Tracking the Hemodynamics in the Evolving False Lumen of Dissected Aorta. Nexus, 1(2), 2024.

Data

The data in this repository consits of computational fluid dynamics results for 3 aneuerysms, as well as the simulated MRI slice data for each aneurysm.

Due to the size of the data it was uploaded to OneDrive.

Code

The Navier-Stokes flow nets base code is required for all other .py files to run, make sure it is included in the same directory.

• Data only
• Warm-start PINN
• Transfer learning warm-start PINN

Cite this work

If you use this code for academic research, you are encouraged to cite the following paper:

@article{Daneker2024,
  author  = {Daneker, Mitchell and Cai, Shengze and Qian, Ying and Myzelev, Eric and Khumbat, Arsh and Li, He and Lu, Lu},
  title   = {Transfer Learning on Physics-Informed Neural Networks for Tracking the Hemodynamics in the Evolving False Lumen of Dissected Aorta},
  journal = {Nexus}
  volume  = {1},
  issue   = {2},
  year    = {2024},
  doi     = {https://doi.org/10.1016/j.ynexs.2024.100016}
}

Questions

To get help on how to use the data or code, simply open an issue in the GitHub "Issues" section.