Deep Learning-Based CT Scan Segmentation may Predict Outcomes in Idiopathic Pulmonary Fibrosis, reveals research
Deep-learning algorithms can effectively segment CT scans of patients with idiopathic pulmonary fibrosis (IPF) and thus offer prognostic information to clinicians, researchers have reported.
According to a team led by Munhunthan Thillai, MD, of the Royal Papworth Hospital in Cambridge, United Kingdom, the data could be used to enhance how IPF patients are followed and treated. The findings were published on November 29, 2023, in the American Journal of Respiratory and Critical Care Medicine.
In the conduction of the study, a group of 446 not-yet-treated IPF patients enrolled in the PROFILE (Prospective Observation of Fibrosis in the Lung Clinical Endpoints) and the authors used deep learning-based CT scan segmentation to test the use of biomarkers (airway, lung, vascular, and fibrosis volumes) for IPF diagnosis and applied them to data received from the enrolled patients. They looked for any correlations between biomarkers and lung function, illness progression, and mortality. Over five years after diagnosis of IPF, the median follow-up was 39.1 months, with a cumulative incidence of death of 277, or 62.1%.
The deep-learning algorithm successfully segmented 97.8% of the CT images, the team reported. It also found the following results:
- Researchers found that higher lung, vascular, and fibrosis volumes were linked to lower five-year survival rates, even after controlling for gender and age.
- Reduced lung capacity and increased fibrosis volume were linked to lower survival rates.
- Lower lung volume, higher vascular volume, and increased fibrosis volume were linked to shorter two-year progression-free survival.
"We demonstrate that CT scans from patients with IPF can be used to train and develop models that can rapidly segment CT scans at scale to produce data on fibrosis, vessel, airway, and lung volumes and that these can predict both progressive disease and mortality," the study's authors concluded.
Reference
https://www.atsjournals.org/doi/10.1164/rccm.202311-2185OC
