Clinical value of magnetic resonance imaging in the early diagnosis of cervical insufficiency in pregnancy

Background Cervical insufficiency (CI) is a leading cause of mid-to-late trimester pregnancy loss, making early and accurate diagnosis critical. While clinical diagnosis primarily relies on transvaginal ultrasound (TVUS), its limitations in visualizing cervical microstructure are recognized. High-resolution magnetic resonance imaging (MRI) offers exceptional soft-tissue contrast, potentially enabling a detailed assessment. This study aimed to explore the descriptive imaging features of MRI in suspected CI.Materials and methods This retrospective study enrolled 77 pregnant women with clinically suspected CI between June 2021 and October 2023. All patients underwent TVUS, and a subset (n = 43) also received MRI. Imaging features, including anatomical visualization quality, endocervical dimensions, and specific MRI findings, were evaluated and described.Results MRI demonstrated significantly superior anatomical visualization compared to TVUS, with satisfaction rates of 100% vs. 67.53% for internal os depiction, 100% vs. 83.11% for external os, and 65.12% vs. 54.55% for amniotic sac protrusion (all p < 0.05). Quantitative measurements revealed a greater endocervical width on MRI (23.47 ± 14.83 mm) compared to ultrasound (12.78 ± 8.41 mm; p = 0.001). Furthermore, MRI detected specific features not visualized by ultrasound, including pericervical defects (88.4%), stromal changes (83.7%), and the “dirty sign” (53.5%). Due to inherent incorporation bias in the reference standard, comparative diagnostic accuracy metrics between modalities were not calculated or reported.Conclusion MRI demonstrates particular promise as a complementary diagnostic tool for cervical insufficiency when ultrasound findings are inconclusive, providing superior anatomical visualization and detecting unique imaging features. The observed association between cervical canal width and CI requires validation in prospective studies designed to mitigate verification and incorporation biases.