Work

Automated quantitative analysis can support the identification of progressive pulmonary fibrosis in fibrotic interstitial lung disease

Year: 2026

Type: article

Abstract: Background Identifying progression in fibrotic interstitial lung disease (f-ILD) is essential for treatment decisions and prognostication. Quantitative CT (qCT) may improve detection of progressive pulmonary fibrosis (PPF). We evaluated whether routine qCT enhances identification of disease progression.Methods We conducted a retrospective observational study of patients presenting to an ILD service over 2 years. Patients with f-ILD who underwent two CT scans within 24 months prior to ILD multidisciplinary team (ILD-MDT) review were included. CTs were initially reported by a mixed pool of radiologists and subsequently reviewed by an expert thoracic radiologist at ILD-MDT. The contextflow ADVANCE Chest CT was used to quantify lung abnormalities.The cohort was divided 70:30 into derivation and validation sets, stratified by ILD-MDT-defined progression. In the derivation cohort, optimal qCT thresholds were defined using Youden’s index and applied to the validation cohort to assess detection of previously unreported progression.Results 127/502 patients met inclusion criteria and were stratified into derivation (n=89) and validation (n=38) cohorts. In the derivation cohort, a 0.39% increase in combined honeycombing and reticulation yielded an AUC of 0.64 for identifying progression.In the validation cohort, this threshold achieved sensitivity and specificity of 0.71 for PPF detection. Although 27/38 (71.1%) patients were initially reported as stable or without comment on progression, 12/27 (44.4%) were later classified as having radiological progression at ILD-MDT. Of these discordant cases, qCT correctly identified progression in 10/12 (83.3%).Discussion This study demonstrates that qCT may assist the non-specialist radiologist in identifying fibrotic disease progression enabling earlier access to appropriate treatment.

Source: BMJ Open Respiratory Research

DOI: 10.1136/bmjresp-2026-004217