AI-Powered CT Scan Analysis Enhances Prognostic Accuracy for Idiopathic Pulmonary Fibrosis, Study Finds
UK: Researchers have reported that deep-learning algorithms can accurately segment computed tomography (CT) scans of patients with idiopathic pulmonary fibrosis (IPF), providing valuable prognostic information for clinicians.
The new data, published in the American Journal of Respiratory and Critical Care Medicine, has the potential to enhance the monitoring and treatment of IPF patients.
Although evidence exists to support the prognostic value of computed tomography (CT) scans in idiopathic pulmonary fibrosis, image-based biomarkers are not commonly employed in clinical practice or trials. Considering this, Muhunthan Thillai, Royal Papworth Hospital, Cambridge, United Kingdom, and colleagues aimed to develop automated imaging biomarkers using deep learning–based segmentation of CT scans.
For this purpose, the researchers developed segmentation processes for four anatomical biomarkers, which were then applied to a unique cohort of treatment-naive patients with IPF from the PROFILE (Prospective Observation of Fibrosis in the Lung Clinical Endpoints) study. These processes were also tested against a separate cohort from the United Kingdom. The study evaluated the relationships among CT biomarkers, lung function, disease progression, and mortality.
The analysis included data from 446 PROFILE patients, with a median follow-up duration of 39.1 months (interquartile range, 18.1–66.4 months). Over five years, the cumulative incidence of death was 277 patients (62.1%).
The study led to the following findings:
- Segmentation was successful on 97.8% of all scans across multiple imaging vendors at slice thicknesses of 0.5–5 mm.
- Lung volume showed the strongest correlation with FVC (r = 0.82) of the four segmentations.
- Lung, vascular, and fibrosis volumes were consistently associated across cohorts with differential 5-year survival, which persisted after adjustment for baseline gender, age, and physiology score.
- Lower lung volume (hazard ratio [HR], 0.98), increased vascular volume (HR, 1.30), and increased fibrosis volume (HR, 1.17) were associated with reduced 2-year progression-free survival in the pooled PROFILE cohort.
- Longitudinally, decreasing lung volume (HR, 3.41) and increasing fibrosis volume (HR, 2.23) were associated with differential survival.
The findings showed that automated models can quickly segment CT scans of IPF, offering both near-term and long-term prognostic information. This capability could be integrated into routine clinical practice or used as critical endpoints in clinical trials.
The team believes these findings could enhance the monitoring of patients with idiopathic pulmonary fibrosis (IPF).
"We show that CT scans from IPF patients can be utilized to train models capable of rapidly and extensively segmenting images to generate data on fibrosis, vessel, airway, and lung volumes. These models can also predict disease progression and mortality," the researchers concluded.
Reference:
Thillai M, Oldham JM, Ruggiero A, Kanavati F, McLellan T, Saini G, Johnson SR, Ble FX, Azim A, Ostridge K, Platt A, Belvisi M, Maher TM, Molyneaux PL. Deep Learning-based Segmentation of Computed Tomography Scans Predicts Disease Progression and Mortality in Idiopathic Pulmonary Fibrosis. Am J Respir Crit Care Med. 2024 Aug 15;210(4):465-472. doi: 10.1164/rccm.202311-2185OC. PMID: 38452227.
