Title:

Accelerating respiratory flow simulations through geometry-aware reduced-order modeling.

Speaker:

Mr. Martin Graffigna

Affiliation:

Georgia Institute of Technology

When:

Thursday, November 20, 2025 at 11:00:00 AM   Add to Calendar

Where:

Boggs Building, Room 3-47

Host:

Shaheen Dewji
shaheen.dewji@gatech.edu

Abstract

Simulating airflow and particle deposition in the human respiratory tract is computationally intensive, limiting its use in time critical or large scale applications. This seminar presents a geometry-aware reduced-order modeling framework that integrates deep autoencoders with Dynamic Mode Decomposition to efficiently represent airway variability and capture flow dynamics. By embedding diverse airway anatomies into a shared latent space and leveraging modal decomposition for temporal evolution, the approach enables drastically faster simulations while preserving physical interpretability. This geometry aware ROM framework offers a path toward rapid, scalable respiratory modeling for applications in inhalation therapy, radiological protection, and personalized dosimetry.

Biography

Martin Graffigna is a Research Engineer at Georgia Tech. He holds a Master of Engineering in Nuclear Engineering from Instituto Balseiro. Martin's research lies at the intersection of high-performance computing, medical imaging, and radiation protection. At the Radiological Engineering, Dosimetry and Detection Laboratory, he develops advanced tools for radiological dose assessment, including deep learning pipelines for rapid biodosimetry and stochastic biokinetics simulators. His work spans automated CFD workflows for inhalation dosimetry, segmentation of 3D airway models from medical images, and urban contaminant dispersion modeling.

Notes

Meet the speaker