Work

Estimating pulmonary arterial pressure from 4D flow images

Year: 2025

Type: article

Abstract: The non-invasive method for estimating the pulmonary arterial pressure (PAP) plays an important role in the screening and diagnosis of pulmonary arterial hypertension (PAH). We aimed to establish a phase-contrast magnetic resonance imaging (PC-MRI) based method to estimate PAP. Through analyzing a patient's morphologic and hemodynamic features, this method could be used to identify PAH, and provide diagnostic and grading information for PAH. We selected 39 study participants, comprising 18 healthy volunteers and 21 patients with PAH. Morphologic and hemodynamic parameters of each participant's pulmonary arteries were obtained from 4D Flow images. Hemodynamic features were performed selected by principal component analysis (PCA). PAH identification model was built with binary logistic regression. Furthermore, A multiple linear regression (MLR) model was developed to estimate PAP, the accuracy of which was evaluated by comparing it with the value measured by right heart catheterization (RHC). PAH identification was achieved with high accuracy, using the features of pulmonary arterial morphology or blood flow velocity (BFV). Compared with RHC, MLR results showed that using pulmonary arterial morphology and BFV features in combination can greatly improve the accuracy of PAP estimation. Our results showed that the mean relative error of PAP estimation in PAH patients could reach <10 %. A highly accurate PC-MRI based method for PAH identification and PAP estimation was successfully established. Using hemodynamic features of the pulmonary artery could improve PAP estimation accuracy, which highlights the importance of hemodynamic evaluation of pulmonary arteries in the screening and diagnosis of PAH. • 4D flow image-based method for pulmonary arterial pressure estimation. • A regression model identifies pulmonary artery hypertension with 100 % accuracy. • A regression model estimates pulmonary arterial pressure with minimum 7.07 % error.

Source: Medicine in Novel Technology and Devices

DOI: 10.1016/j.medntd.2025.100354