Related Experiment Video
Updated: Aug 5, 2026
10:44
Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging
Published on: June 21, 2024
Dynamic Digital Radiography Pulmonary Function Testing: A Machine Learning Lung Study Alternative
Valeria Santibanez1, Thomas J Pisano2, Florence X Doo3
1Department of Pulmonary, The Mount Sinai Hospital, New York, NY.
CHEST Pulmonary
|August 4, 2026
Summary
Machine learning automates dynamic digital radiography (DDR) analysis, creating DDR-based pulmonary function tests (dPFTs). These dPFTs strongly correlate with traditional pulmonary function tests (PFTs), offering a potential alternative diagnostic tool.
Area of Science:
- Pulmonary medicine
- Medical imaging
- Machine learning
Background:
- Pulmonary function tests (PFTs) and chest radiography are standard diagnostic tools for lung disorders.
- These methods offer only static assessments, limiting dynamic lung evaluation.
- Dynamic digital radiography (DDR) captures lung and diaphragm motion but faces adoption barriers due to manual analysis and unclear correlation with PFTs.
Purpose of the Study:
- To develop a machine learning pipeline for automated DDR analysis.
- To assess the correlation between PFT measures and automated DDR-based lung area measurements.
Main Methods:
- A convolutional neural network pipeline was developed to quantify lung areas from DDR images, generating DDR-based PFTs (dPFTs).
- PFT and DDR data were collected from 55 participants.
- Correlations between PFT and dPFT measures were analyzed across normal, obstructive, and restrictive lung physiology.
Main Results:
- Strong, statistically significant correlations were observed between dPFT areas and PFT volumes (e.g., total lung capacity r=0.764, FEV1 r=0.591).
- Automated DDR analysis identified physiological differences using diaphragm curvature and strain analysis.
- Linear regression models successfully derived PFT values from dPFT measurements.
Conclusions:
- DDR-based PFTs (dPFTs) show significant correlation with traditional PFTs, suggesting their utility as a surrogate measure.
- Automated DDR analysis holds potential for integrating DDR as a reliable alternative or supplement to PFTs in clinical practice.
