Abstract
Airway-related quantitative imaging biomarkers are crucial for examination,diagnosis, and prognosis in pulmonary diseases. However, the manual delineationof airway trees remains prohibitively time-consuming. While significant effortshave been made towards enhancing airway modelling, current public-availabledatasets concentrate on lung diseases with moderate morphological variations.The intricate honeycombing patterns present in the lung tissues of fibroticlung disease patients exacerbate the challenges, often leading to variousprediction errors. To address this issue, the 'Airway-Informed Quantitative CTImaging Biomarker for Fibrotic Lung Disease 2023' (AIIB23) competition wasorganized in conjunction with the official 2023 International Conference onMedical Image Computing and Computer Assisted Intervention (MICCAI). The airwaystructures were meticulously annotated by three experienced radiologists.Competitors were encouraged to develop automatic airway segmentation modelswith high robustness and generalization abilities, followed by exploring themost correlated QIB of mortality prediction. A training set of 120high-resolution computerised tomography (HRCT) scans were publicly releasedwith expert annotations and mortality status. The online validation setincorporated 52 HRCT scans from patients with fibrotic lung disease and theoffline test set included 140 cases from fibrosis and COVID-19 patients. Theresults have shown that the capacity of extracting airway trees from patientswith fibrotic lung disease could be enhanced by introducing voxel-wise weightedgeneral union loss and continuity loss. In addition to the competitive imagebiomarkers for prognosis, a strong airway-derived biomarker (Hazard ratio>1.5,p<0.0001) was revealed for survival prognostication compared with existingclinical measurements, clinician assessment and AI-based biomarkers.