Machine Learning Based on Computed Tomography Pulmonary Angiography in Evaluating Pulmonary Artery Pressure in Patients with Pulmonary Hypertension

Aug 5, 2026

Upload by Jiven

Right heart catheterization is the gold standard for assessing pulmonary circulation hemodynamics, especially pulmonary artery pressure (PAP), but its invasive and costly nature limits routine use. A fully automatic machine-learning framework is developed to assess PAP from computed tomography pulmonary angiography (CTPA). Eight pulmonary-artery and heart substructures are segmented and used to build regression and classification models for mPAP and sPAP, evaluated using ICC and AUC.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages19

Reading Time48 min

AccessFree to download

Introduction

Materials and Methods

Data collection and ground-truth labels

Segmentation framework

Regression and classification models

Evaluation metrics

Results

Segmentation performance

Feature–PAP correlations

Model agreement and discrimination

Conclusion

Article Machine Learning Based on Computed Tomography Pulmonary Angiography in Evaluating Pulmonary Artery Pressure in Patients with Pulmonary Hypertension Nan Zhang 1,†, Xin Zhao 2,†, Jie Li 3, Liqun Huang 2, Haotian Li 2, Haiyu Feng 2, Marcos A. Garcia 2, Yunshan Cao 4, Zhonghua Sun 5,* and Senchun Chai 2,* Citation: Zhang, N.; Zhao, X.; Li, J.; Huang, L.; Li, H.; Feng, H.; Garcia, M.A.; Cao, Y.; Sun, Z.; Chai, S. Machine Learning Based on Computed Tomography Pulmonary Angiography in Evaluating Pulmonary Artery Pressure in Patients with Pulmonary Hypertension. J. Clin. Med. 2023, 12, 1297. [https://doi.org/10.3390/jcm12041297](https://doi.org/10.3390/jcm12041297) Academic Editor: Andreas A. Kammerlander Received: 24 November 2022 Revised: 20 January 2023 Accepted: 2 February 2023 Published: 6 February 2023 Copyright: © 2023 by the authors. Li‐ censee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and con‐ ditions of the Creative Commons At‐ tribution (CC BY) license ([https://cre‐](https://cre‐)[ ](https://cre‐)[ativecommons.org/licenses/by/4.0/](ativecommons.org/licenses/by/4.0/)). 1 Department of Radiology, Beijing Anzhen Hospital, Capital Medical University, 2nd Anzhen Road, Chaoyang District, Beijing 100029, China 2 School of Automation, Beijing Institute of Technology, No. 5 Zhongguancun South Street, Haidian District, Beijing 100081, China 3 Department of Pulmonary and Critical Care Medicine, Beijing Anzhen Hospital, Capital Medical University, 2nd Anzhen Road, Chaoyang District, Beijing 100029, China 4 Department of Cardiology, Gansu Provincial Hospital, No. 204, Donggang West Road, Chengguan District, Lanzhou 730099, China 5 Discipline of Medical Radiation Science, Curtin Medical School, Curtin University, Perth 6102, Australia * Correspondence: [z.sun@curtin.edu.au](z.sun@curtin.edu.au) (Z.S.); [chaisc97@163.com](chaisc97@163.com) (S.C.)† These authors contributed equally to this study. Abstract: Background: Right heart catheterization is the gold standard for evaluating hemodynamic parameters of pulmonary circulation, especially pulmonary artery pressure (PAP) for diagnosis of pulmonary hypertension (PH) . However, the invasive and costly nature of RHC limits its wide‐ spread application in daily practice. Purpose: To develop a fully automatic framework for PAP as‐ sessment via machine learning based on computed tomography pulmonary angiography (CTPA) . Materials and Methods: A machine learning model was developed to automatically extract mor‐ phological features of pulmonary artery and the heart on CTPA cases collected between June 2017 and July 2021 based on a single center experience. Patients with PH received CTPA and RHC ex‐ aminations within 1 week. The eight substructures of pulmonary artery and heart were automati‐ cally segmented through our proposed segmentation framework. Eighty percent of patients were used for the training data set and twenty percent for the independent testing data set. PAP param‐ eters, including mPAP, sPAP, dPAP, and TPR, were defined as ground‐truth. A regression model was built to predict PAP parameters and a classification model to separate patients through mPAP and sPAP with cut‐off values of 40 mm Hg and 55 mm Hg in PH patients, respectively. The perfor‐ mances of the regression model and the classification model were evaluated by analyzing the intra‐ class correlation coefficient (ICC) and the area under the receiver operating characteristic curve (AUC). Results: Study participants included 55 patients with PH (men 13; age 47.75 ± 14.87 years) . The average dice score for segmentation increased from 87.3% ± 2.9 to 88.2% ± 2.9 through proposed segmentation framework. After features extraction, some of the AI automatic extractions (AAd, RVd, LAd, and RPAd) achieved good consistency with the manual measurements. The differences between them were not statistically significant (t = 1.222, p = 0.227;

FAQs

Why is pulmonary artery pressure assessment limited in daily practice?

Right heart catheterization provides accurate hemodynamic parameters but is invasive and costly, restricting widespread routine use.

How does the proposed framework predict pulmonary artery pressure from CTPA?

It automatically segments predefined pulmonary-artery and heart substructures from CTPA, then uses extracted features in a regression model for PAP parameters and a classification model for distinguishing patients based on mPAP and sPAP cutoffs.

What metrics were used to evaluate model performance?

Segmentation consistency and model accuracy were assessed using intra-class correlation coefficients (ICC) for agreement and the area under the ROC curve (AUC) for discrimination.

Aug 5, 2026

Upload by Jiven