Development and internal validation of a semiautomated HRCT-based tool for quantification of interstitial lung disease in systemic sclerosis

Abstract

Aim Interstitial lung disease (ILD) is a major cause of morbidity and mortality in systemic sclerosis (SSc), and HRCT disease extent is central to clinical stratification. Visual assessment, however, is subject to interobserver variability. This study aimed to develop and internally validate a semiautomated HRCT-based programme for quantitative assessment of SSc-ILD. Materials and Methods In this retrospective single-centre study, HRCT scans from patients with SSc-ILD were analysed using a semiautomated programme. Pulmonary parenchyma was segmented and classified into ground-glass opacity, reticulation, honeycombing, and normal lung tissue. Intra- and interobserver reproducibility were assessed. Internal validation was performed against expert clinical assessment of ILD extent and pulmonary function test parameters. Results Seventy-eight patients with SSc-ILD were included. Intra- and interobserver reproducibility was excellent, with intraclass correlation coefficients exceeding 0.9 for lung segmentation and lesion quantification. Mean global ILD extent was 21.8% by programme and 17.7% by expert assessment. Programme-derived global extent correlated strongly with expert-assessed extent (r=0.74, p<0.0001). A programme-derived threshold of 19.5% identified expert-assessed extent exceeding 20%, with an area under the receiver operating characteristic curve of 0.885. Conclusion The semiautomated programme provides reproducible quantitative assessment of SSc-ILD on HRCT and shows good agreement with expert clinical evaluation. It may serve as a quantitative adjunct to visual assessment. This was a single-centre study with a relatively small sample size and reliance on a single expert assessor.