Physics-Based Artificial Intelligence (PHAI) Model for Estimating the Diameter of Blood Vessels using PhotoAcoustic PlethysmoGraphy (PAPG)

There is significant interest in continuous noninvasive tracking of arterial diameter for diagnosing and monitoring cardiovascular health. PhotoAcoustic PlethysmoGraphy (PAPG) is a promising technology to enable such tracking and revolutionize vascular diagnostics, guide treatment decisions, and monitor disease progression. This study introduces a novel Physics-based Artificial Intelligence (PHAI) model applied to PAPG signals to estimate blood vessel diameter. PAPG, a non-invasive imaging technique, utilizes laser-induced photoacoustic signals from hemoglobin to capture hemodynamically relevant features. However, the non-linear behavior of PAPG signals, influenced by tissue heterogeneity and blood vessel geometry, poses significant challenges. Our PHAI approach integrates physical modeling with deep learning to address these non-linearities, enabling precise diameter estimation. Validation using tissue-mimicking phantoms demonstrates the model’s accuracy, with a mean absolute error of less than 50 μm for operation frequency of 2.25 MHz.