Model-based iterative reconstruction in CT of paranasal sinuses in cystic fibrosis

AIM: To assess image quality and dose-reduction efficacy of model-based iterative reconstruction (MBIR) in computed tomography (CT) of the paranasal sinus (CTPNS) compared with adaptive statistical iterative reconstruction (ASIR) in cystic fibrosis (CF) patients. MATERIALS AND METHODS: Unenhanced CTPNS studies performed in adult CF patients from 2014 to 2020 were included. MBIR and ASIR were used and compared. Two radiologists assessed the CT images blindly and randomly. Quantitative assessment of noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) was performed in the maxillary sinus, sphenoid body, temporalis, and background air. Qualitative assessment performed included image sharpness, noise and contrast. RESULTS: Thirty-seven MBIR and 45 ASIR CT PNS studies were included. MBIR achieved a 74% effective median dose reduction (0.039 mSv) compared with ASIR (0.147 mSv). Measured noise was significantly lower in all regions using MBIR (p<0.001) with superior SNR (p<0.001). MBIR had higher CNR compared to ASIR (4.567 versus 2.03, p<0.001). MBIR images were sharper and less noisy, with equal contrast. Cohen's weighted kappa value was 0.824 for qualitative analysis, indicating good inter-rater agreement. Both methods had 100% diagnostic acceptability. CONCLUSION: MBIR produces high-quality CTPNS images at a significantly lower dose compared with ASIR. It is the preferred imaging surveillance method in CF patients and may have roles in other patient cohorts. (c) 2021 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.