PO:32:185 | Deep-learning analysis of high-resolution computed tomography images predicts progression and mortality in systemic sclerosis related interstitial lung disease

Valentina Boni¹, Rosa D'Abronzo², Piergiacomo Cerasuolo¹, Lucio Calandriello², Giuseppe Cicchetti², Gerlando Natalello¹, Bruno Iovene⁵, Francesco Varone⁵, Giacomo Sgalla⁵, Luca Richeldi⁵, Anna Rita Larici^2|4, Maria Antonietta D'Agostino^1|3, Silvia Laura Bosello^1|3, Enrico De Lorenzis¹. | ¹Division of Rheumatology and Clinical Immunology Catholic Univeristy of the Sacred Heart, Rome, Italy; ²Department of Diagnostic Imaging, Oncological Radiotherapy and Hematology, Diagnostic Imaging Area, Catholic University, Rome, Italy; ³Department of Geriatric and Orthopaedic Sciences, Catholic Univeristy of the Sacred Heart, Rome, Italy; ⁴Department of Radiological and Hematological Sciences, Section of Radiology, Catholic Univeristy of the Sacred Heart, Rome, Italy; ⁵Division of Pulmonary Medicine, Catholic University of the Sacred Heart, Roma, Italy.

Background. Interstitial lung disease (ILD) is a major complication in systemic sclerosis (SSc) patients, associated with substantial morbidity and mortality. Functional, imaging, and clinical measures of lung involvement could be biased in SSc due to its multiorgan nature and extra-articular involvement (e.g., cardiac, musculoskeletal). Artificial intelligence (AI) reading of high-resolution computed tomography (HRCT) has emerged as a novel tool for the objective and reliable assessment of pulmonary diseases. The aim of this study is to correlate AVIEW measures, an Deep learning based software for HRCT image assessment, with ILD-progression and disease-related mortality in SSc patients.

Materials and Methods. The AVIEW software (Coreline Soft, South Korea) was employed to analyze HRCT images from a cohort of consecutive SSc-ILD patients at baseline and after 24±3 months. Quantitative analyses included lung volume, texture, airways, and vascular anatomy. Baseline metrics were assessed for their association with ILD progression, defined by clinical, functional, and imaging criteria based on the INBUILD study parameters over 24 months. Furthermore, changes in AVIEW-derived measurements between consecutive HRCT evaluations over the 24-month period were analyzed for their association with SSc-related mortality during the subsequent 36 months. All absolute measurements were normalized to body surface area.

Results. A total of 146 HRCT scans from 73 SSc-ILD patients were assessed (mean age 58.4±14.3 years; male 16.4%; diffuse skin variant 49.3%). Thirty-one patients (42.4%) experienced ILD progression, which was predicted at baseline by higher percentages of ground glass opacities (GGO) (p=0.05) and reticulation (p=0.05), higher subpleural vessel volumes (p=0.017), and a tendency toward larger distal airways (p=0.066). Serial evaluations demonstrated that INBUILD progression was associated with a reduction in the percentage of normal lung (p=0.044) and absolute volumes (p=0.009), without significant changes in reticulation, GGO, vessels, or airways when considered individually. Twelve patients died due to SSc within 36 months following the second HRCT evaluation. Patients in the upper quartile for changes in reticular score and airway volume exhibited a higher mortality risk, independent of INBUILD progression (reticular score: OR 3.30, 95% CI 1.03–10.61, p=0.045; airway volume: OR 3.37, 95% CI 1.08–10.51, p=0.036)

Conclusions. Deep learning-based assessment in SSc-ILD identified distinct modifications and prognostic significance in lung anatomical components, offering potential improvements in patient evaluation and stratification beyond conventional clinical tools.