Machine Learning May Help Distinguish COPD From Asthma Using CT Features

By Jeff Craven, MD /alert Contributor
Save to PDF By

Researchers using machine learning have identified several disease-related features highlighted in CT that can help distinguish chronic obstructive pulmonary disease (COPD) from asthma, according to recent research published in the European Respiratory Journal.

The researchers evaluated 93 CT imaging features of 95 patients with either COPD or asthma to determine which CT features were optimal in a machine learning model for identifying the differences between COPD and asthma. The patients had similar baseline ages and forced expiratory volume in 1 second scores. 

Among the CT imaging features extracted included total airway count (TAC), percentage of low-attenuating area below −950 HU (LAA950), low-attenuation cluster total hole count, the estimated airway wall thickness in an idealized airway with an inner perimeter of 10 mm (Pi10), measurements for five segmental airways (RB1, RB4, RB10, LB1, and LB10) that included airway inner/outer perimeters/areas and wall thickness, and averages for the five segmental airways.

Using machine learning, the investigators applied hybrid feature selection to determine the optimal number of CT imaging features, while they used a support vector machine learning algorithm to differentiate between COPD and asthma. The CT features from the model results showed an accuracy percentage of 80% and an F1 score of 81%. 

The algorithm selected seven key CT features: LAA950, outer and inner airway perimeter, TAC, outer and inner airway area RB1, and low-attenuation cluster total hole count. The researchers also estimated CT features in a model that only evaluated airway features and found an accuracy of 66% and an F1 score of 68%. Within this model, the top CT features included inner airway area RB1, outer airway area LB1, outer and inner airway perimeter, Pi10, TAC, airway wall thickness RB1 and TAC LB10. 

The researchers noted that airway remodeling/loss and intraluminal mucus occlusion could be responsible for or partially explain CT TAC reduction in asthma and may be a treatable feature that can be detected using CT imaging. They also acknowledged that future studies were required to determine if CT evidence of airway mucus occlusion, or reduced CT TAC, could be utilized to identify patients with improved responses to mucus clearing treatments.

“Imaging has the potential to provide a greater understanding of the underlying structural changes that contribute to airflow limitation in asthma and COPD patients, and therefore identification of these structural changes may allow for a more trait-based approach to treatment, particularly in those with features of both asthma and COPD,” Amir Moslemi, PhD, of the department of physics at Ryerson University in Toronto, and colleagues said. 


Disclosures: Some authors declared financial ties to drugmakers. See full study for details.

Photo Credit: Getty Images

© 2023 /alert® unless otherwise noted. All rights reserved.
Reproduction in whole or in part without permission is prohibited.
Privacy Policy | Terms of Use | Editorial Policy | Advertising Policy