What the study evaluated
This study is a narrative review of published research on the use of deep learning and artificial intelligence for detecting and monitoring cystic fibrosis–related lung changes on computed tomography (CT), particularly high-resolution CT. The review analyses existing deep learning architectures, quantitative imaging approaches, and their potential integration into clinical practice.
Study results in clinical practice
The reviewed studies consistently show that deep learning enables automated and reproducible quantification of key cystic fibrosis features such as bronchiectasis, mucus impaction, airway wall thickening, and air trapping. In clinical practice, AI has the potential to reduce radiologists’ workload, improve consistency compared to visual scoring systems, and support earlier and more precise monitoring of disease progression. However, widespread clinical adoption depends on further validation, standardisation, and regulatory approval.
Key numbers
Type of study: literature review
Databases searched: PubMed
Studies included: 19 relevant publications
Primary imaging modality: CT / HRCT
Focus: automated detection and quantitative assessment of CF-related lung changes
Cystic fibrosis (CF) is a genetic disease caused by mutations in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene. This disorder causes a wide range of clinical complications, primarily affecting the respiratory and digestive systems and extending its impact to other physiological areas. Early detection and careful monitoring are paramount to mitigate disease progression and improve the quality of life of individuals with CF. Computed tomography (CT), particularly high-resolution CT (HRCT), has become a key diagnostic method for detecting pulmonary manifestations of CF. However, manual analysis of CT images requires a high level of expertise and is time consuming. The combination of artificial intelligence (AI) and deep learning with CT imaging predicts significant advances in CF detection. Deep learning, a subset of AI, uses neural networks to analyse complex morphological patterns indicative of disease from large datasets. This review traces the journey from the earliest attempts to use artificial intelligence in CF detection to recent advances made using deep learning algorithms. By exploring various deep learning architectures and their integration into clinical practice, this review illuminates the potential of these new technologies to revolutionize CF detection using CT imaging. Automated and accurate analysis enabled by deep learning aims to reduce the diagnostic burden on radiologists, speed up the diagnostic process and pave the way for timely and personalized therapeutic interventions, which is in line with the ultimate goal of improving patient care.
