Holistic segmentation of CT structural alterations with deep learning has recently been described in cystic fibrosis (CF), allowing the measurement of Normalized Volumes of Airway Abnormalities (NOVAA-CT) as an automated quantitative outcome. Clinical evaluations are still needed, including longitudinal evaluations. The aim of the study is to assess the clinical validity of NOVAA-CT in CF patients with and without CFTR modulators.

The study was retrospective and included three independent cohorts between 2014 and 2023. Thirty-four CF patients undergoing Elexacaftor/Tezacaftor/Ivacaftor (ETI), and twenty patients undergoing Corticosteroids for allergic broncho-pulmonary aspergillosis (ABPA) at a single Institution composed the ETI and ABPA groups, respectively. Fifty-four CF patients from geographically-distinct Institutions composed an External group. All patients had completed CT and pulmonary function test (PFT) with measurement of the Forced Expiratory Volume in 1-second percentage predicted (FEV1%p), along with a second assessment at one year in case of ETI or ABPA treatment. NOVAA-CT included quantification of bronchiectasis, peribronchial thickening, bronchial mucus, bronchiolar mucus, collapse/consolidation, and total abnormal volumes (TAV). Two observers also scored the visual Bhalla score.

. All correlations between NOVAA-CT and either the Bhalla score or FEV1%p were significant (rho>0.7; p<0.001) in all ETI, ABPA, and External groups (n=34; n=20; n=54, respectively), at all timepoints. In both ETI and ABPA groups, there was significant improvement in bronchial and bronchiolar mucus volumes (p<0.01) after ETI and corticosteroid treatment, respectively. An additional reversibility in normalized bronchiectasis volume was quantified in the ETI group (p<0.001). The variation in normalized TAV significantly correlated to the variation in FEV1%p in the ETI group and ABPA group. The reproducibility of NOVAA-CT was almost perfect (ICC>0.99).

NOVAA-CT automatically quantifies structural abnormalities’ volumes over an entire lung, enabling to reproducibly monitor the disease severity and characterize therapeutic effects on lung structure at CT.