A Machine Learning Framework for Arterial Health Classification using Ultrasound Vibrometry Data

Presenter: Wilkins Aquino

The Center for Computational and Digital Health Innovation is proud to invite you to join our monthly seminar series. This month, we are featuring Duke faculty member Wilkins Aquino. Abstract below:
Arterial stiffness is a critical indicator of cardiovascular health. To noninvasively assess arterial health, we utilize arterial dispersion ultrasound vibrometry (ADUV), where an acoustic radiation force (ARF) generates propagating waves in the arterial wall, measured through ultrasound. A persistent challenge in this domain is developing a fast and reliable method to map these wave measurements to arterial health metrics. In response, we propose an end-to-end classification framework that directly links ADUV data to arterial health assessment. Key contributions of our approach include effective handling of high-dimensional signals, optimal feature selection for interpretability, and the training of efficient classifiers. We demonstrate the effectiveness of our ADUV-based machine learning framework across cohorts, including healthy individuals, patients with atherosclerotic cardiovascular disease (ASCVD), and those with cardiovascular risk factors (CVDRF).

Contact

Marnie Rhoads
marnie.rhoads@duke.edu

More Information