A groundbreaking study published in Radiology: Cardiothoracic Imaging reveals that a deep learning model can accurately diagnose and stage chronic obstructive pulmonary disease (COPD) using just a single inhalation lung CT scan. This approach eliminates the need for the dual CT protocol traditionally used in COPD diagnostics, potentially making the process more accessible and cost-effective.
COPD, the third leading cause of death globally according to the World Health Organization, is a progressive lung disease characterized by symptoms such as shortness of breath and fatigue. Currently, spirometry tests are the standard for diagnosis, supplemented by dual CT imaging that includes inspiratory and expiratory scans to assess lung function.
However, dual CT protocols are not widely adopted due to challenges such as the need for additional training and the difficulty some elderly patients face in holding their breath during exhalation imaging.
Researchers, led by Dr. Kyle A. Hasenstab of San Diego State University, developed a convolutional neural network (CNN) model that leverages a single inspiratory CT scan alongside clinical data to diagnose and stage COPD. Using data from 8,893 patients with a history of smoking, the model accurately predicted spirometry measurements and classified the severity of COPD based on the Global Initiative for Obstructive Lung Disease (GOLD) staging system.
The study found that the CNN model’s performance matched that of traditional methods requiring both inhalation and exhalation CT scans. Adding clinical data further improved the accuracy of the predictions.
“Reducing the protocol to a single inspiratory CT scan can enhance the accessibility of COPD diagnostics while lowering costs, minimizing patient discomfort, and reducing radiation exposure,” Dr. Hasenstab stated.
As reported by news-medical.net,this innovation holds promise for broader adoption of COPD diagnostics across healthcare settings, providing a streamlined, patient-friendly solution for identifying and managing the disease.