Search

IntroductionSegmentation of lung structures in medical imaging is crucial for the application of automated post-processing steps on lung diseases like cystic fibrosis (CF). Recently, machine learning methods, particularly neural networks, have demonstrated remarkable improvements, often outperforming conventional segmentation methods. Nonetheless, challenges still remain when attempting to segment various imaging modalities and diseases, especially when the visual characteristics of pathologic findings significantly deviate from healthy tissue.MethodsOur study focuses on imaging of pediatric CF patients [mean age, standard deviation (7.50 ± 4.6)], utilizing deep learning-based methods for automated lung segmentation from chest magnetic resonance imaging (MRI). A total of 165 standardized annual surveillance MRI scans from 84 patients with CF were segmented using the nnU-Net framework. Patient cases represented a range of disease severities and ages. The nnU-Net was trained and evaluated on three MRI sequences (BLADE, VIBE, and HASTE), which are highly relevant for the evaluation of CF induced lung changes. We utilized 40 cases for training per sequence, and tested with 15 cases per sequence, using the Sørensen-Dice-Score, Pearson’s correlation coefficient (r), a segmentation questionnaire, and slice-based analysis.ResultsThe results demonstrated a high level of segmentation performance across all sequences, with only minor differences observed in the mean Dice coefficient: BLADE (0.96 ± 0.05), VIBE (0.96 ± 0.04), and HASTE (0.95 ± 0.05). Additionally, the segmentation quality was consistent across different disease severities, patient ages, and sizes. Manual evaluation identified specific challenges, such as incomplete segmentations near the diaphragm and dorsal regions. Validation on a separate, external dataset of nine toddlers (2–24 months) demonstrated generalizability of the trained model achieving a Dice coefficient of 0.85 ± 0.03.Discussion and conclusionOverall, our study demonstrates the feasibility and effectiveness of using nnU-Net for automated segmentation of lung halves in pediatric CF patients, showing promising directions for advanced image analysis techniques to assist in clinical decision-making and monitoring of CF lung disease progression. Despite these achievements, further improvements are needed to address specific segmentation challenges and enhance generalizability.

Introduction: Chronic rhinosinusitis (CRS) usually presents with nasal congestion, rhinorrhea and anosmia impacts quality of life in cystic fibrosis (CF). Especially mucopyoceles pathognomonic for CRS in CF may cause complications such as spread of infection. Previous studies using magnetic resonance imaging (MRI) demonstrated early onset and progression of CRS from infancy to school age in patients with CF, and mid-term improvements of CRS in preschool and school-age children with CF treated with lumacaftor/ivacaftor for at least 2 months. However, long-term data on treatment effects on paranasal sinus abnomalities in preschool and school-age children with CF are lacking.Methods: 39 children with CF homozygous for F508del (mean age at baseline MRI 5.9 ± 3.0 years, range 1-12 years) underwent MRI before (MRI1) and about 7 months after starting lumacaftor/ivacaftor and then annually (median 3 follow-up MRI, range 1–4) (MRI2-4). MRI were evaluated using the previously evaluated CRS-MRI score with excellent inter-reader agreement. For intraindividual analysis ANOVA mixed-effects analysis including Geisser-Greenhouse correction and Fisher’s exact test, and for interindividual group analysis Mann-Whitney test were used.Results: The CRS-MRI sum score at baseline was similar in children starting lumacaftor/ivacaftor in school age and children starting therapy at preschool age (34.6 ± 5.2 vs.32.9 ± 7.8, p = 0.847). Mucopyoceles were the dominant abnormality in both, especially in maxillary sinus (65% and 55%, respectively). In children starting therapy in school age the CRS-MRI sum score decreased longitudinally from MRI1 to MRI2 (−2.1 ± 3.5, p < 0.05), MRI3 (−3.0 ± 3.7, p < 0.01) and MRI4 (−3.6 ± 4.7, p < 0.01), mainly due to a decrease in the mucopyoceles subscore (−1.0 ± 1.5, p = 0.059; −1.2 ± 2.0, p < 0.05; −1.6 ± 1.8, p < 0.01; and −2.6 ± 2.8, p = 0.417, respectively). In children starting lumacaftor/ivacaftor in preschool age, the CRS-MRI sum score remained stable under therapy over all three follow-up MRI (0.6 ± 3.3, p = 0.520; 2.4 ± 7.6, p = 0.994; 2.1 ± 10.5, p > 0.999 and −0.5 ± 0.5, p = 0.740; respectively).Conclusion: Longitudinal paranasal sinus MRI shows improvements in paranasal sinus abnormalities in children with CF starting lumacaftor/ivacaftor therapy at school age. Further, MRI detects a prevention of an increase in paranasal sinus abnormalities in children with CF starting lumacaftor/ivacaftor therapy at preschool age. Our data support the role of MRI for comprehensive non-invasive therapy and disease monitoring of paranasal sinus abnormalities in children with CF.