P352 Comparison of Radiation Dose, Image Quality, and Diagnostic Performance between Hybrid Iterative Reconstruction and Deep Learning Reconstruction for CT Enterography of IBD

Abstract

Abstract Background CT enterography (CTE) is recommended as one of the preferred methods for patients with inflammatory bowel disease (IBD), especially in primary hospitals that lack MRI equipment. However, the potential risk of radiation derived from using CTE during the life-long follow-up is an important issue for these patients. This study aimed to compare the radiation dose, image quality, and diagnostic performance of CTE for IBD using hybrid iterative reconstruction (HIR) and deep learning reconstruction (DLR). Methods We prospectively collected data on low dose CTE reconstructed by HIR and DLR in patients with IBD between February 2023 and September 2023, and retrospectively enrolled IBD patients who underwent conventional dose CTE reconstructed by HIR between June 2019 and June 2022 as controls. The CTE examinations were performed within 2 weeks of bowel histologic evaluation (the reference standard of bowel inflammation). The objective and subjective image quality and the image features were evaluated by two IBD radiologists. The radiation dose, image quality, and diagnostic performance was compared between the two dose protocols. Results The preliminary study included 36 patients in the low dose group with HIR and DLR and 40 patients in the conventional dose group with HIR. There were no significant differences in demographic data of patients between the two groups (all P>0.05). The sensitivity, specificity, and diagnostic accuracy for detecting active inflammation were 71.4% (43.3%), 83.3% (80.0%), and 75.0% (52.5%) for the low dose group with DLR (HIR), respectively, and 84.4%, 62.5%, and 80% for the conventional dose HIR group. Low dose group with DLR resulted in significantly increased objective and subjective image quality and diagnostic confidence for active inflammation compared to low dose group with HIR (all P<0.05), with no inferiority to conventional dose group with HIR (all P>0.05). The dichotomous inter-reader reliability (k) for the entire group was 0.88. Low dose group resulted in a significant reduction in radiation exposure (mean [± SD], 2.20 ± 0.29 mSv) compared to conventional dose group (4.85 ± 1.96 mSv; P<0.001). Conclusion Low dose CTE with DLR significantly improved image quality compared to low dose CTE with HIR, and maintained similarly high diagnostic performance as conventional dose CTE with HIR. Importantly, the low dose CTE with DLR significantly reduced the radiation dose, improving the safety of CTE application in the life-long follow-up of patients with IBD.