COMPANY
Thirona
Healthcare technology company based in Nijmegen, the Netherlands
Founded 2014Nijmegen, the Netherlands
AI quantification of bronchial dimensions shows stronger correlation with lung function compared to manual measurements in bronchiectasis patients
Company: Thirona Product: LungQ
Automated method of bronchus and artery dimension measurement in an adult bronchiectasis population
ERJ Open Research, 2024
Abstract
Aim
Bronchiectasis (BE) is a disease defined by irreversible dilatation of the airway. Computed tomography (CT) plays an important role in the detection and quantification of BE. The aim of this study was three-fold: 1) to assess bronchus-artery (BA) dimensions using fully automated software in a cohort of BE disease patients; 2) to compare BA dimensions with semi-quantitative BEST-CT (Bronchiectasis Scoring Technique for CT) scores for BE and bronchial wall thickening; and 3) to explore the structure-function relationship between BA-method lumen dimensions and spirometry outcomes.
Methods
Baseline CTs of BE patients who participated in a clinical trial were collected retrospectively. CTs were analysed manually with the BEST-CT scoring system and automatically using LungQ (v.2.1.0.1, Thirona, The Netherlands), which measures the following BA dimensions: diameters of bronchial outer wall (Bout), bronchial inner wall (Bin) and artery (A), and bronchial wall thickness (Bwt) and computes BA ratios (Bout/A and Bin/A) to assess bronchial widening. To assess bronchial wall thickness, we used the Bwt/A ratio and the ratio between the bronchus wall area (Bwa) and the area defined by the outer airway (Boa) (Bwa/Boa).
Results
In total, 65 patients and 16 900 BA pairs were analysed by the automated BA method. The median (range) percentage of BA pairs defined as widened was 69 (55-84)% per CT using a cut-off value of 1.5 for Bout/A, and 53 (42-65)% of bronchial wall were thickened using a cut-off value of 0.14 for Bwt/A. BA dimensions were correlated with comparable outcomes for the BEST-CT scoring method with a correlation coefficient varying between 0.21 to 0.51. The major CT BA determinants of airflow obstruction were bronchial wall thickness (p=0.001) and a narrower bronchial inner diameter (p=0.003).
Conclusion
The automated BA method, which is an accurate and sensitive tool, demonstrates a stronger correlation between visual and automated assessment and lung function when using a higher cut-off value to define bronchiectasis.