CT-based radiomics for assessing lung function (decline) in SSc-ILD patients: a promising approach

<bold>Introduction:</bold> Systemic sclerosis (SSc) is a rare connective tissue disease often accompanied by rapidly evolving interstitial lung disease (SSc-ILD). Given the prominent use of CT imaging in SSc-ILD diagnosis and follow-up, reliable image quantification can aid in effective patient management. <bold>Aim:</bold> This study aims at evaluating the association of CT-based radiomics with conventional Pulmonary Function Tests (PFT). <bold>Methods:</bold> We retrospectively analysed baseline and follow-up CT-scans of 85 SSc-ILD patients, collected at the University Hospital of Liège. Follow-up scans were available for 65 patients. Radiomic features, including Total Lung Volume (TLV) and normalized High Attenuated Volume (HAV), were extracted from the automatically segmented lung region (Radiomics.bio, Belgium). We assessed the association between radiomic features and PFT using the Pearson correlation coefficient (ρ). <bold>Results:</bold> CT-based TLV positively correlated with TLC (ρ=0.79, p<0.01), FVC (ρ=0.81, p<0.01) and FEV1 (ρ=0.77, p<0.01). Several shape features like lung surface and compactness showed similar correlation behaviour. HAV negatively correlated with FVC (ρ=-0.48, p<0.01) and FEV1 (ρ=-0.47, p<0.01). Different intensity and texture features correlated with DLCO (GLSZM_ZE: ρ=-0.53, p<0.01; P90 intensity: ρ=-0.52, p<0.01; intensity skewness: ρ=0.5, p<0.01). Correlations were also found in longitudinal changes, for example ΔAsphericity correlated with ΔTLC (ρ=0.48, p<0.01), ΔFVC (ρ=0.43, p<0.01) and ΔFEV1 (ρ=0.41, p<0.01). <bold>Conclusion:</bold> CT-based radiomics extracted from the lung region is a promising approach to assess lung function (decline) and could potentially be used to identify progressive disease in SSc-ILD.